Menu engineering: the four dishes you must act on

Menu engineering is a simple method for deciding what to do about every dish on a menu. It looks at two things for each dish: how popular it is, and how much money it makes each time it sells. Put those two measures together and every dish falls into one of four groups, each of which calls for a different action. Some dishes should be protected, some re-costed, some repositioned and some replaced.

The method has been around for decades, and it remains one of the most useful tools a venue has, because it turns a vague sense that "the menu needs looking at" into a short list of specific changes. It also corrects a common mistake: judging dishes by popularity alone, or by food cost percentage alone, when what matters is how much each dish contributes in total.

This guide explains menu engineering step by step: the data you need, how to measure popularity and gross profit per dish, how to sort dishes into the four quadrants, and the single action each quadrant calls for. It works through an illustrative menu of eight mains, then covers the dishes that break the rules, how often to repeat the exercise and the mistakes that make it misleading.

Menu engineering is a way of analysing a menu by comparing each dish's popularity with its profitability. The idea, often credited to work by Michael Kasavana and Donald Smith in the early 1980s, is described in the general entry on menu engineering. It uses two measures: the dish's share of sales within its section, and its contribution margin, which is the selling price minus the cost of the ingredients.

Each measure is compared with a benchmark for the section. A dish is popular if its share of sales is above the popularity benchmark, and profitable if its contribution margin is above the section's average. That produces four combinations, and four groups of dishes.

The power of the method is that it judges each dish by the money it actually contributes, not by its food cost percentage. A dish with a high food cost percentage can still make more money per plate than a cheaper dish with a low one. Menu item profitability is about pounds, dirhams or dollars per plate, multiplied by how many plates sell.

Menu engineering needs three numbers for each dish over the same period, usually a month or a menu cycle:

Units sold come from your sales records. Selling prices come from the menu. Food cost comes from recipe costing: each ingredient's quantity in the dish multiplied by its cost per unit, including garnishes and sauces. Our guide to recipe costing covers how to cost a dish properly.

Analyse one section at a time. Mains should be compared with mains, starters with starters, desserts with desserts. Comparing a dessert with a steak tells you nothing useful, because guests are not choosing between them.

Popularity in menu engineering is measured as each dish's share of the units sold in its section. If a section sold 1,600 mains in a month and one dish sold 420 of them, its share is about 26 per cent.

The benchmark for popularity is usually set at 70 per cent of an equal share. With eight dishes, an equal share would be 12.5 per cent each; 70 per cent of that is 8.75 per cent. Any dish with a share above 8.75 per cent counts as popular. The 70 per cent rule allows for the fact that, in a real menu, sales are never evenly spread, and a dish does not need an equal share to be pulling its weight.

Measure popularity from the menu mix over a full period, not from a single busy night. Our guide to menu mix covers how to read it by category and by hour, and how to spot a dish whose share is changing.

Gross profit per dish, or contribution margin, is the selling price excluding tax minus the food cost of one portion. A dish that sells for 24.00 and costs 8.40 to make has a contribution margin of 15.60. The idea is the same as the general concept of contribution margin: what each unit contributes towards fixed costs and profit.

The benchmark is the section's weighted average contribution margin: the total contribution margin for all dishes in the section, divided by the total units sold. Weighting by units means that popular dishes count for more, which reflects how the section actually makes its money.

A dish with a contribution margin above the average counts as profitable. A dish below it counts as less profitable, even if its food cost percentage looks good. This is the point most often missed: a cheap dish with a low food cost percentage can still contribute less per plate than an expensive dish with a higher one.

Combining popularity and profitability gives four groups of dishes. The names vary, but the most common are these:

Each group calls for a different response. The value of menu engineering is that it tells you which response each dish needs, rather than treating the whole menu the same way.

Stars are the dishes that carry the menu. The single action they call for is protection. Keep their quality consistent, keep them in prominent positions on the menu, make sure the kitchen can always make them, and resist the temptation to change them for the sake of change.

Be careful with price. A star can often absorb a modest price rise, because guests order it for reasons other than price, but test any rise carefully and watch its share afterwards. A star that loses its popularity because of a careless price rise becomes a puzzle.

Watch stars for drift. Portion sizes creep up, garnishes become more generous, and suppliers raise prices. Re-cost stars regularly, so their margin does not erode while everyone assumes they are fine.

Workhorses are popular dishes with below-average margins. The single action they call for is re-costing: finding ways to raise their contribution margin without losing their popularity. Removing a workhorse is rarely the answer, because guests come for it.

Start with the plate. Is the portion larger than it needs to be? Is there an expensive garnish that guests would not miss? Could a costly component be replaced with an equally good, cheaper one? Small changes to the recipe can raise the margin without guests noticing.

Then consider price. A small price rise on a dish that guests love is often accepted, especially if it brings the dish in line with similar dishes elsewhere. You can also pair a workhorse with a higher-margin side or add-on, so the order as a whole contributes more. Our guide to menu pricing covers the options.

Puzzles make good money per plate but do not sell enough. The single action they call for is repositioning: making them more visible and more appealing, so more guests order them.

Look at where the dish sits on the menu and how it is described. A puzzle buried in the middle of a long section, with a dull description, may simply be overlooked. Move it to a more noticeable position, give it a clearer and more appetising description, highlight it, or ask servers to recommend it. Our guide to menu design covers placement and descriptions.

If repositioning does not work, the dish may be priced too high for what guests expect, or it may simply not suit your guests. A small price reduction can sometimes turn a puzzle into a star; if not, consider whether it belongs on the menu at all.

Dogs neither sell well nor make good money. The single action they call for is replacement: take them off the menu and use the space for a dish with more potential. A shorter menu is easier to read, easier to cook and usually sells better.

There are exceptions, and they matter. A dish may be a dog by the numbers but serve a purpose: the only vegan main, the dish a regular orders every week, or the plain option that lets a group with one cautious eater choose your restaurant. Removing it may cost more in lost tables than it saves.

If a dog stays for a reason, write the reason down and review it. Then improve the dish where you can: re-cost it so it contributes more, and make it as good as it can be, so the guests who need it are glad it is there.

Here is an illustrative month of eight mains, to show the method. The dishes and figures are examples, not a real venue's menu. Contribution margin is the price minus food cost.

The section sold 1,600 mains. The popularity benchmark is 70 per cent of an equal share of 12.5 per cent, or 8.75 per cent, which is 140 units. The total contribution margin is 20,596, so the weighted average margin is 20,596 divided by 1,600, or about 12.87.

That gives the four groups:

The actions follow directly. Protect the stars: keep the sea bream, steak and chicken consistent and prominent. Re-cost the workhorses: look at the burger's portion and garnish, consider a small price rise, and pair the fish and chips with a higher-margin side. Reposition the puzzles: move the risotto to a more visible spot with a better description, and ask servers to recommend the lamb for two.

The vegan curry is the interesting case. By the numbers it is a dog, but it is the only vegan main. Removing it could cost tables where one guest is vegan and the group would otherwise choose another restaurant. The better action is to improve it: re-cost it, refresh the recipe, describe it better, and review it again next month.

Notice also what the analysis does not say. It does not say the steak is a bad dish because its food cost percentage is around 41 per cent, the highest on the menu. It says the steak is a star, because it sells well and contributes 16.50 per plate. Judging by food cost percentage alone would have pointed in the wrong direction.

Menu engineering is a guide, not a law, and some dishes need judgement. New dishes need time before they can be judged; a dish in its first month may look like a puzzle simply because guests have not discovered it yet. Seasonal dishes may sell strongly for a few weeks and then fade.

Some dishes earn their place indirectly. A signature dish that defines the venue, a dish that brings in large groups, or a dish that encourages guests to order a bottle of wine may contribute more than its own numbers show. Record these reasons, and check them against the evidence from time to time.

And some dishes interact. Removing a popular but low-margin dish may push guests to a star, or it may push them to another restaurant. Change one or two dishes at a time, and watch the menu mix to see where demand goes.

Classic menu engineering looks at ingredients only. It does not count the time a dish takes to prepare, the skill it needs or the pressure it puts on a busy station. Two dishes with the same contribution margin can be very different to run: one plated in a minute from prepared components, the other needing ten minutes of attention on the grill at the busiest point of the night.

Bring that into your judgement. A star that takes a long time on the bottleneck station may limit how many covers the kitchen can serve, and a workhorse that is quick and easy may be worth more than its margin suggests. Our guide to pacing covers covers why the busiest station sets the kitchen's limit.

You do not need a precise figure for labour per dish. A simple note of which dishes are slow or difficult, beside the menu engineering results, is enough to stop the analysis recommending a menu the kitchen cannot cook on a Saturday.

Menu engineering produces decisions, and the floor team carries many of them out. If a puzzle is to be repositioned, servers need to know why it matters and how to recommend it. If a workhorse has been re-costed, the kitchen needs to know the new portion. If a dish has been removed, everyone needs to know what to suggest instead.

Explain the reasons in the briefing, briefly and without jargon. "We want more people to try the risotto, because it is one of our best dishes and guests who order it love it" is more useful than a quadrant chart. Ask servers what guests say about the dishes in question; they often know why a puzzle is not selling.

Run the analysis at least every time the menu changes, and ideally monthly or quarterly for menus that stay the same for longer. Costs change, popularity shifts with the seasons, and a dish that was a star in winter may be a puzzle in summer.

Keep each analysis, so you can compare them over time. A dish that moves from star to workhorse over three months is telling you that its costs have risen or its price has fallen behind. A puzzle that becomes a star after repositioning tells you the change worked.

The first analysis takes the longest, because recipes have to be costed and data gathered. Later analyses are quicker, especially if your sales system and recipe costs are kept up to date.

With the Analytics module, servearound's menu report shows every dish that sold, with its units, revenue and share, and its gross margin once its cost is set. Categories show where the revenue and margin sit, top sellers are listed beside every dish that sold, and the menu's own funnel shows how far guests get, from opening the menu to checking out. Those are the two numbers menu engineering needs, popularity and margin, read from the checks the dishes were actually on.

When the analysis tells you to act, the menu is where you act. Dishes live in one library behind every menu, so a new description, a highlight or a price change made once appears on the QR code menu, the screens in the room and the printed menu together. See the menus module and Analytics.

No. A spreadsheet with units sold, prices and food costs is enough. Software makes it faster and keeps the data current.

Yes, as a separate analysis. Compare drinks with drinks, and cocktails with cocktails.

Start by costing your best-selling dishes, then work through the rest. Even rough costs are better than none, but the more accurate they are, the better the decisions.

No. What matters is the contribution margin per plate and how many plates sell. A dish with a high food cost percentage can still be a star.

One or two per section, so you can see the effect of each change on the menu mix.

Analyse them separately, or over the period they run. Specials can be a good way to test a dish before adding it to the main menu.

Yes, though with only four or five dishes per section the benchmarks move a lot with each sale. Use a longer period, and lean more on judgement when a dish sits close to a benchmark.

Menu engineering compares each dish's popularity with its contribution margin, within its section, and sorts dishes into four groups. Stars are popular and profitable: protect them. Workhorses are popular but less profitable: re-cost them. Puzzles are profitable but less popular: reposition them. Dogs are neither: replace them, or justify them.

Use accurate recipe costs and a full period of sales, judge dishes by gross profit per dish rather than food cost percentage, and combine the analysis with judgement about the dishes that earn their place in other ways. Change one or two dishes at a time, watch the menu mix, and repeat the analysis whenever menu pricing or the menu itself changes.