What is menu engineering, in one page

What is menu engineering? It is a simple way of judging every dish on a menu by two questions at once: how often do guests order it, and how much does it earn each time? Plot every dish on a chart with popularity on one side and profit on the other, and the menu sorts itself into four groups. Each group calls for a different action: protect, reprice, promote or replace.

The idea has been around for decades and is often wrapped in consulting language: matrices, contribution margins, quadrant analysis. Underneath, it is common sense. A dish that sells a lot and earns a lot is a star. A dish that sells a lot but earns little needs its price or cost looked at. A dish that earns a lot but rarely sells needs help to be noticed. A dish that does neither is taking space from something better.

This guide explains menu engineering in plain language: the idea in one page, the two axes, the four categories and what to do with each, and the arithmetic behind them. It covers menu mix, menu item profitability and gross profit per dish, the mistakes that make the method mislead, and how often to do it.

Here is menu engineering in one page:

Everything else in this guide is detail.

Menu engineering as a named method is usually traced to work by hospitality academics in the early 1980s, who suggested judging dishes by popularity and contribution margin together rather than by food cost percentage alone. The names they gave the four groups, stars, plowhorses, puzzles and dogs, are still used today, along with plainer alternatives.

The method has been refined many times since, but the core idea has stayed the same.

Many venues judge dishes by food cost percentage: the cost of the ingredients as a share of the price. A dish with a low percentage looks good; one with a high percentage looks bad. But a percentage hides the money. A steak with a 40 per cent food cost may earn far more per sale than a pasta with 20 per cent, because the steak's price is so much higher.

Menu engineering looks at the money each dish earns, gross profit per dish, and how often it sells. Together, they show what each dish contributes to the venue's profit.

The first axis is popularity: how often a dish is ordered compared with the other dishes in its section. In a section of ten mains, each would sell about a tenth of the section's total if all were equally popular. A dish that sells much more than that is popular; one that sells much less is not.

A common rule treats a dish as popular if its share of the section is at least 70 per cent of an equal share. In a section of ten, that is 7 per cent of the section's sales or more.

The second axis is profit: gross profit per dish, the price after tax minus the plate cost. Compare each dish's gross profit with the average gross profit of its section, weighted by how many of each were sold. A dish above the average is high profit; one below it is low.

Use the money, not the percentage. The question is how much each sale leaves to pay for everything else, which is the dish's contribution margin in the language of accounting.

The two axes divide the menu into four categories:

The names do not matter. What matters is the action each group calls for.

Stars sell well and earn well. They are the backbone of the menu. Protect them: keep their quality and portion consistent, keep them in prominent places on the menu, make sure the team knows them well, and be careful with price rises, which can be tested in small steps.

Do not change a star without a good reason. Many venues have damaged a star by "improving" it.

Plowhorses sell well but earn little per sale. Guests like them, but the venue makes less on each than on its stars. The usual actions are to raise the price a little, if the dish can bear it, or to lower the plate cost: a cheaper garnish, a slightly smaller portion of the most expensive ingredient, or a side that costs less.

Make small changes and watch sales. A plowhorse that stops selling after a price rise has become a dog.

Puzzles earn well but rarely sell. Something is stopping guests ordering them: the description, the position on the menu, the name, the price, or simply that the team never mentions them. The usual actions are to move them to a better position, rewrite the description, have the team recommend them, or feature them as a special.

If a puzzle still does not sell after promotion, consider whether it belongs on the menu.

Dogs neither sell well nor earn well. They take up space on the menu, in the kitchen and in the stock room. The usual actions are to rework them, with a new recipe, a new price or a new description, or to remove them and make room for something better.

Some dogs stay for good reasons: a vegetarian option the menu needs, a dish for children, a regular's favourite. Keep them consciously, not by accident.

To engineer a section of the menu:

Here is an illustrative section of four mains, over a month. The figures are examples.

With 870 mains sold, an equal share is about 218; the 70 per cent threshold is about 152. The weighted average gross profit is about 15. Actions: protect the steak, test a small price rise on the chicken, move the lamb up the menu with a better description, and rework or replace the risotto.

Menu mix is the share of sales each dish takes within its section or the whole menu. It is the popularity axis of menu engineering, and it is useful on its own: it shows what guests actually choose, which is often different from what the kitchen thinks they choose.

Our guide to menu mix: what your guests order, not what you sell covers how to read it.

Menu item profitability is the profit axis: how much each dish earns. It depends on accurate plate costs, which depend on accurate recipes and current supplier prices. A dish costed a year ago with old prices may look profitable when it is not.

Our guide to recipe costing covers how to cost a dish properly, from the box price to the plate.

Gross profit per dish is the price after tax minus the plate cost: the money each sale leaves to pay for labour, rent and everything else. It is the most useful single number for judging a dish, because it is in money rather than a percentage. Our guide to what is gross profit on a dish explains why.

Engineer each section of the menu separately: starters with starters, mains with mains, desserts with desserts. Comparing a dessert with a steak tells you nothing useful, because guests do not choose between them. Within a section, guests are choosing one dish over another, and that choice is what the method reads.

Drinks can be engineered too, by category: wines by the glass, cocktails, beers.

Use a period long enough to smooth out odd days but short enough to reflect the current menu: usually a month, or the life of a menu if it changes more often. Exclude dishes that were only on for part of the period, or adjust for the days they were available.

Avoid periods with unusual trade, such as a festive month with set menus, unless that is what you want to study.

Menu engineering can mislead if the inputs are wrong. Plate costs that are out of date put dishes in the wrong category. Sales that include set menu items at special prices distort popularity. Sections that mix very different dishes, such as small plates and large mains, compare unlike with unlike. A single month after a menu change may not yet show settled habits.

Check the inputs before acting on the outputs.

Menu engineering does not tell you why a dish is popular or slow. It does not show whether guests liked it, whether it was sent back, whether it slows the kitchen, or whether it brings guests in who then order other things. A dish that looks like a dog may be the reason a group of regulars visits.

Use the categories to ask questions, then answer them with what you know about the guests and the kitchen.

Some dishes take far more kitchen time than others. A dish that earns well but takes a long time to prepare at the peak may cost more in slower service than it earns. Consider kitchen effort alongside the four categories, especially for puzzles you plan to promote.

If promoting a dish would overload one station, think again.

Set menus complicate the picture. A dish sold inside a set menu at a fixed price per guest does not have its own price on that check, so its popularity counts but its profit is harder to read. Engineer the à la carte menu separately from set menus, or count set menu dishes in popularity only.

Engineer the set menu itself as a whole: its price per guest against its total plate cost, and how often it sells compared with the à la carte choices.

Specials can be engineered too, but they move quickly. A special that sells strongly and earns well may deserve a place on the main menu; one that sells poorly despite the team's recommendations is telling you something. Keep a simple record of each special's sales and gross profit.

Specials are also a low-risk way to test a dish before it joins the menu, and to see where it would land in the four groups. A few weeks as a special often says more than any amount of discussion in the kitchen.

Many of the actions in menu engineering are about the menu itself: where a dish sits, how it is described, whether it has a box or a picture, how its price is shown. Guests tend to read certain places on a menu first, and dishes there sell more.

Our guide to menu design for reading, not for framing covers layout, and our guide to menu engineering: the four dishes you must act on goes further into the actions.

Servers influence what guests order. A dish the team recommends sells more; one they never mention sells less. When you decide to promote a puzzle, tell the team why, let them taste it, and ask them to suggest it. When you remove a dog, explain the change.

The method works best when the floor and the kitchen understand it.

Do it every few months, and after every significant menu change, once the new menu has settled for a few weeks. Some venues do a quick version monthly for their most important section, usually the mains, and a full version each quarter.

Keep a record of each round: the categories, the actions and the results. Over time, it shows which changes worked.

After acting, measure the result: did the plowhorse's sales hold after the price rise? Did the puzzle sell more after it moved? Did the section's total gross profit rise? Compare the same length of period before and after, and allow for seasonal changes.

The goal is not to move every dish into the top group, but to raise the total profit from the menu while keeping guests happy.

A small menu of eight or ten dishes can still be engineered, but with fewer dishes the categories are rougher, and one busy weekend can move a dish from one group to another. Use a longer period, and treat the results as prompts for discussion rather than verdicts.

In a group, engineer each venue's menu separately, because guests and costs differ. A dish that is a star in one venue may be a dog in another. Comparing the results across venues can show where a shared dish needs a local price, recipe or position.

Here is an illustrative quarter for a venue with twelve mains. The details are examples.

At the next review, total gross profit from mains rose, with mains sold roughly steady.

In servearound, menu analytics plots every dish on a menu engineering matrix: popularity against revenue, or against contribution margin once dish costs are set, with the four groups named top earners, popular but thin margin, profitable but slow, and weak sellers, and a hint for each. Each dish can be opened for a closer look.

The menu report shows every dish that sold, with units, checks, per check, revenue, share, unit cost and margin, how much of the revenue has a cost behind it, and what diners looked at against what they ordered. See Menus and Analytics.

A way of judging every dish by how often it sells and how much it earns, sorting the menu into four groups with an action for each.

High sales and high profit, high sales and low profit, low sales and high profit, and low sales and low profit.

Because a percentage hides the money. Gross profit per dish shows what each sale actually earns.

By each dish's share of its section's sales, often compared with 70 per cent of an equal share.

By gross profit per dish, compared with the section's average, weighted by sales.

Every few months, and after each menu change once the new menu has settled.

Not always. Some stay for good reasons, such as a vegetarian option or a regular's favourite.

Yes, by category: wines by the glass, cocktails, beers and so on.

The head chef and the general manager together: the chef knows the costs and the kitchen effort, the manager knows the sales and the guests. Neither should change the menu on the matrix alone.

What is menu engineering? Judging every dish by two questions at once: how often it sells, the menu mix, and how much it earns, its gross profit per dish. Plot the dishes in each section, and they fall into four groups: protect the stars, reprice or recost the plowhorses, promote the puzzles, and rework or remove the dogs.

Use current plate costs for accurate menu item profitability, engineer each section separately, choose a sensible period, and remember what the method does not show, such as kitchen effort and why guests choose a dish. Act, measure the result, and repeat every few months.