The menu engineering matrix, with a real menu

The menu engineering matrix is one of the most useful tools a venue has, and one of the most often misused. The idea is simple: plot every dish by how popular it is and how much it earns, split the chart into four quadrants, and act on each dish according to where it lands. In practice, most attempts go wrong at the same step: drawing the two lines that create the quadrants.

Draw the lines in the wrong place and dishes land in the wrong quadrants. A solid earner is marked for removal; a dish that is quietly losing money is protected. The chart looks convincing, and the decisions it leads to are wrong.

This guide works through the menu engineering matrix with an example menu, section by section, and shows exactly how to draw the two lines, how to read each quadrant, and where the common mistakes hide. Every figure in it is an example, from an invented menu, chosen to make the arithmetic easy to follow.

The menu engineering matrix compares dishes on two measures at once. The first is popularity: how often each dish sells compared with the others in its section. The second is profitability: how much cash each dish contributes after its ingredient cost, its contribution margin.

Plot every dish on those two measures and draw a line across each, and the chart splits into four quadrants:

The method was set out in the early 1980s by Michael Kasavana and Donald Smith, and Wikipedia's article on menu engineering covers its background. Our guide to menu engineering, and the four dishes you must act on covers the actions in more depth. This guide is about plotting the chart correctly.

The first rule is to plot each section of the menu separately. Starters compete with starters, mains with mains, desserts with desserts. Putting them all on one chart makes no sense: a dessert will almost always earn less cash than a steak, and comparing them tells you nothing useful.

Take an example menu of twenty dishes: six starters, ten mains and four desserts. That is three separate charts, each with its own lines. This guide works through the ten mains in detail, because they carry the most money, but the same steps apply to every section.

Leave out specials that ran for only part of the period, set menus priced as a whole, and sides, unless sides are a large part of the business. They distort the averages that the lines depend on.

You need two numbers for every dish over the same period: how many were sold, and the contribution margin of each one. A month is usually a good period: long enough to smooth out a strange night, short enough to reflect the current menu.

Units sold come from the sales report. The contribution margin comes from your costing: the selling price, net of tax, minus the ingredient cost of one portion. Our guide to recipe costing, one dish from box price to plate covers how to cost a dish properly.

Use cash margin, not food cost percentage. A dish with a low food cost percentage can still earn little cash if its price is low; a dish with a higher percentage can earn more cash if its price is high. The matrix is about cash, and our guide on why gross profit per dish beats food cost percentage explains why.

Here are the ten example mains for one month, with units sold, selling price, ingredient cost and contribution margin:

In total, the section sold 1,000 mains in the month. That total, and the margins, are what the two lines are built from.

The popularity line separates dishes that sell well from those that do not. The obvious way to draw it is at the average: with ten dishes, each would sell 10% of the section if sales were even, so any dish above 10% is popular.

The classic method does something slightly different, and it matters. It sets the line at 70% of that even share. With ten dishes, the even share is 10%, and 70% of that is 7%. So a dish is counted as popular if it sells at least 7% of the section, which here is 70 of the 1,000 mains.

The reason is practical. Sales are never even, and on any menu a few dishes sell far more than the rest. If the line sat at the plain average, most dishes would fall below it and be marked unpopular, even though many of them sell perfectly respectably. The 70% rule draws the line where a dish genuinely starts to struggle.

The profitability line separates dishes that earn more than the section's typical dish from those that earn less. Here, too, the obvious method is slightly wrong.

The obvious method is the simple average of the ten margins: add them up and divide by ten. The better method is the weighted average: the total margin the section earned, divided by the total number of dishes sold. That reflects what a typical plate actually earns, given how many of each dish were sold.

In the example, the total margin is the margin of each dish multiplied by its units, all added together: 13,217.50. Divided by 1,000 mains, the weighted average margin is about 13.22. That is where the profitability line sits. A dish earning more than 13.22 a plate is above the line; a dish earning less is below it.

With both lines drawn, each dish falls into a quadrant:

Three stars, four plowhorses, two puzzles and a dog. Already, some of the results deserve a second look, which is the next step.

Dishes close to a line deserve care, because a small change in the data would move them. In the example, roast chicken and the risotto earn 13.00, a fraction below the 13.22 line. Pork belly sells exactly 70, right on the popularity line.

Treat these as borderline rather than firmly in a quadrant. Roast chicken is not really a problem dish; it is a popular dish that earns almost exactly the section's typical margin, and a small price rise or a small saving on its cost would make it a star. The same is true of the risotto.

A useful habit is to mark any dish within a few per cent of a line as borderline, and to look at it over a second period before acting. Decisions based on a dish that sits right on a line are decisions based on noise.

To see why the lines matter, draw the popularity line at the plain average instead of the 70% rule: at 10%, or 100 mains, rather than 70.

The risotto, at 90, drops below the line and, with its margin just under the other line, becomes a dog instead of a plowhorse. Sea bass, at 80, drops from star to puzzle. Pork belly, at 70, drops from star to puzzle. Suddenly the section has only one star, the steak, four puzzles and two dogs, and a manager reading the chart might conclude that the menu is full of dishes that need promoting or removing, when in fact three of them are doing well.

The profitability line can mislead in the same way if it is drawn from food cost percentages rather than cash margins, or from a simple average when a few dishes sell far more than the rest. Getting the lines right is not a technicality. It decides which dishes you act on.

Stars are the dishes that carry the section: popular and profitable. In the example, the steak, the sea bass and the pork belly.

The action for stars is to protect them. Keep their quality and portion consistent, keep them in strong positions on the menu, and be cautious about changing them. A star can usually carry a modest price rise if costs rise, because guests choose it for reasons beyond price.

Watch them over time. A star that slips towards the popularity line may be losing appeal, perhaps because a competitor offers something similar or because the dish has drifted from its original standard. Our guide to menu mix: what your guests order, not what you sell covers how to follow those shifts.

Plowhorses are popular but earn less than the typical dish. In the example, the burger, fish and chips, the roast chicken and the risotto. They sell well, which means guests like them, so removing them is rarely the answer.

The actions are about margin. Look first at cost: a cheaper supplier, a smaller or better-controlled portion of the expensive component, a different side. Then look at price: a popular dish can often carry a small increase without losing many sales. Sometimes both: a burger whose cheese portion has grown can earn its margin back with a reminder about the spec and a small price rise.

Be careful with plowhorses that guests come for. If the burger is why many guests choose the venue, its margin is only part of its value.

Puzzles earn well but do not sell as often as they should. In the example, the lamb shoulder and the duck breast. Each plate is valuable; there are just not enough of them.

The actions are about selling. Is the dish in a good position on the menu, or buried in the middle of a long section? Is its description appealing and clear? Do servers recommend it? Is it priced so high that guests hesitate? Sometimes a better description, a better position and a word from the server are enough to move a puzzle towards the stars.

Look at whether guests even notice the dish. If guests open the dish on a digital menu but rarely order it, the description or the price may be putting them off; if they rarely open it at all, its position on the menu is the more likely problem.

Dogs sell little and earn little. In the example, the vegan curry. On a pure reading of the chart, it is the candidate for removal.

But removing a dish from the menu needs more than the chart. The vegan curry may be the only vegan main, and removing it would leave guests with that need with nothing to eat, which costs the venue far more than the curry's modest margin. In that case, the action is to rework it: a better recipe, a better description, a price that earns more, rather than removal.

Where a dog has no such reason to stay, replace it with something that might do better. Our guide to removing a dish from the menu without a riot covers how to do it without upsetting the guests who did order it.

The same steps apply to the starters and the desserts, each with its own lines. With six starters, the even share is about 16.7% and the 70% popularity line about 11.7%; with four desserts, the even share is 25% and the line 17.5%. The profitability line for each section is that section's own weighted average margin.

Expect the results to look different. Desserts often have fewer dishes and more even sales, so fewer of them fall below the popularity line. Starters often include a couple of cheap, popular dishes that land as plowhorses. Neither is a problem; it is simply what those sections look like.

Do not compare quadrants across sections. A star among the desserts earns far less cash than a plowhorse among the mains, and that is fine. Each chart tells you how a dish performs against its real competitors.

The matrix is a powerful first view, but it uses only two measures. Menu item profitability has other dimensions the chart does not show.

Kitchen time is one. A dish that earns a good margin but takes a long time to prepare, or ties up a busy station at the peak, may earn less per minute of kitchen time than a simpler dish. Waste is another: a dish whose ingredients are used nowhere else may carry hidden waste that its costing does not show.

Our guide to menu item profitability, dish by dish covers how to rank dishes by margin per minute of kitchen time as well as by cash margin. Use the matrix to see the pattern, and those wider measures before making a final decision on any dish.

Redraw the matrix regularly: monthly for a busy venue, quarterly for a quieter one, and always after a menu change. Each new chart shows whether the actions from the last one worked: whether the plowhorse's price rise held its sales, whether the puzzle's new description lifted its orders.

Keep the old charts. A dish that moves from puzzle to star over three months is evidence that the selling changes worked. A star that slides towards the line over several charts is an early warning.

Keep the method the same each time, with the same period length, the same treatment of specials and the same 70% rule, so that the charts can be compared fairly.

In servearound, the menu report gives the popularity side of the matrix: units sold, checks, revenue and share of sales for every dish and category over any dates, which is the menu mix the popularity line is drawn from. From the menu guests open on their phones, it also shows how often each dish was looked at against how often it was ordered, which helps tell a puzzle that guests never see from one they see and pass over.

servearound does not hold dish costs today, so the contribution margins, the profitability line and the chart itself come from your own costing sheet, with the units and revenue taken from the report. Inventory tracking, for the stock behind the menu, is coming. See the menus module and Analytics.

A chart that plots dishes by popularity and contribution margin, split into four quadrants: stars, plowhorses, puzzles and dogs, each with its own action.

No. Plot each section separately, because starters, mains and desserts earn very different margins and compete only with each other.

The classic method sets it at 70% of the even share: with ten dishes, the even share is 10%, so the line is at 7% of the section's sales.

At the weighted average contribution margin: the section's total margin divided by the total number of dishes sold.

Because the matrix is about cash. A cheap dish with a low food cost percentage can still earn less cash per plate than a dearer dish with a higher percentage.

Treat it as borderline and look at another period before acting. Small changes in the data move it from one quadrant to another.

No. A dog that serves a dietary need, such as the only vegan main, should usually be reworked rather than removed.

Monthly for a busy venue, quarterly for a quieter one, and after every menu change, using the same method each time.

Kitchen time, waste and the role a dish plays in bringing guests in. Check those before making a final decision on any dish.

The menu engineering matrix plots each dish by popularity and contribution margin and splits the chart into stars, plowhorses, puzzles and dogs. Its value depends on the two lines. Plot each section separately, draw the popularity line at 70% of the even share, and draw the profitability line at the weighted average cash margin, not an average of food cost percentages.

Look again at dishes near the lines before acting, protect the stars, work on the cost and price of plowhorses, work on how puzzles are sold, and rework or replace dogs, unless a dog serves a need the menu must meet. Redraw the chart regularly with the same method, and check kitchen time and waste before any final decision.