Restaurant menu engineering worksheet explained: what the matrix actually does
A restaurant manager watches the dinner rush roll through on a Friday night. The ribeye sells out, the salmon sits, and the chicken special is reordered three times even though the margin is thin. By Sunday, the question is the same one every operator eventually faces: which dishes are quietly making money, which are quietly losing it, and which deserve a rewrite, a reprice, or a quiet retirement. A restaurant menu engineering worksheet is the tool that turns that instinct into a one-page decision.
The worksheet is the working surface behind menu engineering, a method restaurant operators have used since the late 1970s to classify menu items by both profitability and popularity. Rather than relying on a chef’s gut or a server’s hunch, the worksheet forces every dish on the menu through the same two filters: how much profit each plate actually contributes, and how often guests actually order it. The result is a four-quadrant view of the menu that makes the next decision obvious, or at least defensible.
This guide walks through the worksheet end to end. It explains where the numbers come from, how the four classic categories are defined, how to score each dish, how to read the matrix once it is filled in, and how to turn the picture into menu changes. The aim is practical: by the end, a reader should be able to build a working restaurant menu engineering worksheet for their own operation without buying software or hiring a consultant.
Where menu engineering came from and why the worksheet still matters
Menu engineering as a discipline is usually credited to the work done at the Cornell Hotel and Restaurant Administration school in the late 1970s and early 1980s. The original framework asked a simple question: if a restaurant knows exactly which items sell and exactly which items make money, why are decisions about the menu still being made on instinct. The answer the researchers proposed was a structured classification, paired with a worksheet that any operator could fill in by hand.
The core insight has aged well. Most restaurants track food cost as a percentage of revenue, but percentage of food cost alone hides a great deal. A dish with a 28 percent food cost on a popular pasta can contribute less total profit than a dish with a 35 percent food cost on a premium steak, simply because of price and volume. The worksheet captures both axes at once, which is why it has survived the move from paper ledgers to point-of-sale reports to modern restaurant analytics platforms.
The format has also branched. The original matrix uses contribution margin and sales mix. A popular variant called menu profitability engineering uses item profit and popularity in dollars rather than in mix percentages. A more recent variant, menu engineering 2.0, swaps simple popularity for a more nuanced guest satisfaction signal drawn from comment cards or online reviews. The worksheet is the same shape in each case: rows for dishes, columns for the two scoring inputs, and a four-quadrant plot at the end.
For independent operators, the worksheet is still useful because it does not require a subscription. A spreadsheet, a calculator, and one solid week of sales data are enough. For chain operators, the same logic scales across hundreds of locations, which is why the menu engineering worksheet is still part of training at many culinary and hospitality programs.
The two numbers every menu item needs
Before the worksheet can be filled in, two values have to be calculated for every dish on the menu. Neither is complicated, and both are the kind of number a kitchen should already be tracking in some form.
Contribution margin per dish
Contribution margin is what is left of the selling price after the food cost of the dish is removed. It is sometimes called item profit or plate profit, and it answers the question: if a guest orders this, how many dollars does it actually put toward rent, labor, and the rest of the operating overhead.
The formula is straightforward:
Contribution margin = menu price minus plate food cost
Plate food cost is the cost of the ingredients that go onto the plate, not the cost of the labor to cook it, and not the cost of the dishware. Most kitchens build plate cost from a recipe that lists every component in measurable units, multiplied by the current cost of each ingredient from invoices. A 6 ounce portion of salmon at $1.20 per ounce, a half cup of rice at $0.18, and a tablespoon of citrus butter at $0.30 puts the plate food cost at $7.68. If the salmon sells for $26, the contribution margin is $18.32.
Sales mix or popularity
The second number is popularity, usually expressed as sales mix. Sales mix is the share of total menu items sold that each dish represents. A pasta that sells 40 times in a week out of 400 total menu items has a 10 percent sales mix. The math is also simple:
Sales mix percent = units of this dish sold divided by total menu items sold, multiplied by 100
Some operators prefer to count dollars rather than units. That choice changes the matrix but not the logic. Counting units rewards dishes that bring guests in the door, which is useful when the operator wants to drive traffic. Counting dollars rewards dishes that bring revenue in, which is useful when capacity is tight. For most independent restaurants, counting units is the more honest measure, because a high-priced dish that sells only twice a night can look artificially important in dollar terms.
How to set the two thresholds that split the matrix
The four quadrants of the matrix only make sense once a midpoint is set on each axis. On the horizontal axis, the midpoint is the average contribution margin across the menu. On the vertical axis, the midpoint is the average sales mix, also across the menu. Dishes above the midpoint on either axis are above average on that measure; dishes below are below average.
Setting the midpoints from the menu’s own averages matters because the threshold is relative, not absolute. A neighborhood cafe and a fine dining tasting menu can both use the same worksheet because the midpoints move with the menu. The cafe might split the matrix at a $4 contribution margin and a 4 percent sales mix. The tasting room might split at $14 and 3 percent. The interpretation is the same in both cases.
Some teams prefer to use median rather than mean, since menu sales tend to skew toward a few popular items. A handful of bestsellers can pull the average up, which leaves most of the menu sitting below the line by default. Median treats every dish as one vote. Either is defensible. What matters is that the rule is written down before the worksheet is filled in, so the operator is not tempted to move the line to put a favorite dish in a flattering quadrant.
The classic four-quadrant matrix and what each box means
Once the two midpoints are drawn, the menu drops into four boxes. The vocabulary varies slightly by author, but the categories are stable across the literature and across the many spreadsheet templates in circulation.
Stars: high margin, high popularity
Stars are the dishes every operator wishes they had more of. They earn above-average contribution margin per plate and they sell above their share of the menu. In a casual restaurant, that might be a blackened mahi sandwich. In a steakhouse, the bone-in ribeye. Stars are the items to protect, feature, and engineer around.
The practical work for stars is to make sure they stay in stock, stay well positioned on the menu, and stay well described. Stars also justify small price increases, because guests are already choosing them for reasons beyond price. A star that has been quietly subsidized by an overgenerous portion is a candidate to be re-costed, since the margin may be flattered by waste that the kitchen has not yet noticed.
Plowhorses: low margin, high popularity
Plowhorses are the workhorses of the menu. Guests order them often, but the contribution margin is below average. A plowhorse is not a failure. It is usually a sign that a dish is delivering traffic, not dollars. The typical examples are big salads, bone-in wings, and value entrées at family restaurants.
The action on a plowhorse is almost always engineering, not deletion. Small portion adjustments, ingredient swaps, and price nudges of a dollar or two are usually enough to push a plowhorse toward star territory. Removing a plowhorse is risky, because the item is the reason some guests walk in. The first question to ask of a plowhorse is whether its current portion makes sense, and the second is whether its current price reflects what the kitchen actually spends to produce it.
Puzzles: high margin, low popularity
Puzzles are profitable per plate but do not sell. They are often premium items or unfamiliar preparations: lamb shank, whole grilled branzino, a vegetarian entrée that costs little to produce but that guests pass over for safer choices. Puzzles are the most variable category, because low sales can be caused by price, by description, by position, by server behavior, or by simple unfamiliarity.
The first move on a puzzle is to ask why the dish is not moving. Sometimes the answer is that it sits in a low-attention corner of the menu. Sometimes the description undersells it. Sometimes servers have quietly learned to talk guests out of it. The cheapest fix is usually repositioning or rewriting the description, and the worksheet can be paired with a quick read of server notes to tell the two cases apart. If a puzzle still does not sell after a better spot on the menu and a sharper description, the dish may be a candidate for a quiet retirement.
Dogs: low margin, low popularity
Dogs are the dishes the menu can probably do without. They do not earn strong margin and they do not bring guests in. In some cases a dog is a legacy item that a long-time regular still orders. In other cases it is a dish that the chef added for personal reasons and that the menu no longer needs. Either way, dogs are the first candidates for removal, and a clean worksheet is the best way to start that conversation with the kitchen team without making it personal.
Some dogs are worth a rescue attempt. A dish that is a dog because of a recent ingredient cost spike may be a star again once the cost normalizes. A dish that is a dog because of a tired description may move into puzzle territory with a rewrite. The worksheet is most useful when it is filled in over several periods, so the operator can see whether a dog is a one-week blip or a long-term pattern.
A blank restaurant menu engineering worksheet at a glance
The worksheet has the same shape regardless of which software it lives in. A simple version uses four columns, with one row per dish. The four columns are: menu price, plate food cost, contribution margin, and units sold. Two derived columns handle the categorization: a sales mix percentage, and a quadrant label. The end of the sheet holds the averages that draw the two threshold lines on the matrix.
| Column | What it holds | How it is calculated |
|---|---|---|
| Menu price | The price printed on the menu | Taken directly from the current menu |
| Plate food cost | Cost of all ingredients for one serving | Recipe build times current ingredient cost |
| Contribution margin | Dollars left after food cost | Menu price minus plate food cost |
| Units sold | Plates sold in the chosen period | POS count for the same period |
| Sales mix | Share of total units this dish represents | Units sold divided by total units, multiplied by 100 |
| Quadrant | Star, plowhorse, puzzle, or dog | Set by margin and mix against the menu averages |
Below the line items, the worksheet holds the two averages that drive the matrix: the mean contribution margin across the whole menu, and the mean sales mix across the whole menu. Those two numbers are the crosshairs of the chart.
A worked example with twelve menu items
To make the math concrete, consider a small bistro with twelve menu items, all priced between $14 and $32. The kitchen pulls a week of point-of-sale data, and the manager builds a quick worksheet. The numbers below are an illustration, not a real menu, but they are the kind of shape a real worksheet tends to take.
| Dish | Menu price | Plate cost | Margin | Units | Mix % | Quadrant |
|---|---|---|---|---|---|---|
| Cast iron chicken | $22.00 | $6.40 | $15.60 | 118 | 14.0% | Star |
| Butcher burger | $18.00 | $5.90 | $12.10 | 132 | 15.6% | Plowhorse |
| Citrus salmon | $28.00 | $9.10 | $18.90 | 71 | 8.4% | Star |
| Wild mushroom pasta | $21.00 | $5.20 | $15.80 | 58 | 6.9% | Star |
| Short rib ragu | $26.00 | $8.40 | $17.60 | 34 | 4.0% | Puzzle |
| Roasted cauliflower | $17.00 | $3.80 | $13.20 | 52 | 6.2% | Star |
| Grilled hangar steak | $32.00 | $12.30 | $19.70 | 29 | 3.4% | Puzzle |
| House meatballs | $16.00 | $5.60 | $10.40 | 96 | 11.4% | Plowhorse |
| Falafel plate | $15.00 | $3.20 | $11.80 | 48 | 5.7% | Plowhorse |
| Seared scallops | $30.00 | $11.80 | $18.20 | 14 | 1.7% | Puzzle |
| French onion soup | $11.00 | $2.90 | $8.10 | 74 | 8.8% | Plowhorse |
| Seared tofu bowl | $16.00 | $4.10 | $11.90 | 19 | 2.3% | Dog |
| Total / average | — | — | $14.44 | 845 | 100% | — |
With the average margin at about $14.44 and the average mix at 100 divided by 12, or roughly 8.3 percent, the matrix splits as the table above suggests. The seared tofu bowl sits below average on both axes and is the only dog. The four stars and the four plowhorses each get a different kind of attention. The menu already leans healthy, which is a useful signal of its own: stars are concentrated in the middle of the price range, while puzzles cluster at the high end. That is the typical pattern for a menu that has been allowed to drift toward premium items without enough support from the front of the house.
How to read the matrix once it is filled in
Reading the matrix is mostly a matter of looking for clusters and gaps. A menu with a single star and a dozen dogs is telling a very different story from a menu with six stars and two dogs, and the actions follow from the shape.
Where the stars cluster
Stars that cluster in one part of the menu, say the pasta section or the grill section, suggest that the kitchen has a real strength and that the menu should lean into it. Stars that scatter across categories with no clear pattern suggest that the menu is a collection of one-off dishes rather than a coherent program, which is worth thinking about during the next menu refresh.
Where the plowhorses live
Plowhorses in a tight cluster usually point to a category that the kitchen is undercharging. A row of plowhorse sandwiches, for example, often means the same bread, the same proteins, and the same sauces are being sold at similar price points, and a small, coordinated price and portion review is the cleanest fix. Plowhorses scattered across categories usually point to a portion standard that has slipped over time, which is also fixable but harder to spot without the worksheet.
Where the puzzles hide
Puzzles often hide in corners of the menu that the guest rarely sees, like the second page of a long menu or the bottom of a small type section. A puzzle that sits in a high-visibility slot and still does not sell is a much harder problem than a puzzle that has been quietly placed out of the way. The first puzzle is usually a sign of a mismatch between dish and audience. The second puzzle is usually a sign of a placement problem.
Where the dogs gather
One or two dogs on a menu is normal. A menu that is half dogs is a menu that has not had a real review in a long time. The worksheet is the right place to start the conversation, but the actual removal should usually wait until the kitchen has had a chance to propose replacements. Removing three dogs and adding one star is a better trade than removing three dogs and adding three more puzzles.
Turning the worksheet into menu decisions
The matrix is a diagnostic, not a prescription. The next step is to convert each quadrant into a small list of actions and to rank those actions by how much money they are likely to move.
Actions for stars
- Keep the dish prominently placed, ideally above the fold of a printed menu or near the top of a tablet menu.
- Refresh the description periodically so it does not become invisible through familiarity.
- Re-cost the dish every quarter to make sure the margin is still real, especially after supplier price changes.
- Test small price increases of one to two dollars if the dish is consistently selling out.
- Use the dish as an anchor for the section, and place complementary items nearby so guests see them at the same time.
Actions for plowhorses
- Adjust portion size first, since most plowhorses are portion-heavy rather than poorly priced.
- Reduce the cost of one or two ingredients through a supplier review, without changing the guest experience.
- Move the price up by a small amount, ideally in a way that keeps the dish within the same price tier on the menu.
- Consider a side upcharge to recover margin on high-cost garnishes or accompaniments.
- Use the dish as a traffic driver in marketing, since the popularity is an asset, not a liability.
Actions for puzzles
- Rewrite the description to focus on flavor, origin, and texture rather than on technique or unfamiliar terms.
- Move the dish into a higher-attention slot, even temporarily, to test whether placement was the issue.
- Brief the service team so servers can speak about the dish with confidence and without overselling.
- Test the dish as a special for a defined window to gather a clean signal of demand.
- If the puzzle still does not move, retire it and replace it with a stronger candidate from the chef’s pipeline.
Actions for dogs
- Remove the dish from the printed menu first, and test whether anyone orders it off-menu.
- Keep a written record of the dish in a recipe archive, so it can return if conditions change.
- Use the freed menu slot to elevate a star or to test a new puzzle.
- Communicate the change to the front-of-house team so servers are not surprised when guests ask.
How long a worksheet period should be
One week is the most common period for a restaurant menu engineering worksheet, because it captures both a busy night and a quiet night and gives a representative sample of the week. Shorter periods exaggerate the effect of a single busy Saturday. Longer periods blur in seasonal shifts and price changes. Two to four weeks is reasonable for a menu that is fairly stable, especially if the operator wants to smooth out a holiday weekend or a local event.
The period should match the rhythm of the menu, not the calendar. A seasonal menu that changes every three months deserves a worksheet in the middle of each season, plus one a few weeks after the launch to see whether the new items are settling in. A year-round menu deserves a worksheet at least once per quarter, with an extra pass after any major ingredient price change.
The same rule applies to source data. Sales data should come from the same period as the recipe costs. A worksheet that uses last week’s sales with last quarter’s ingredient prices will quietly misclassify anything whose cost has moved, which is a lot of dishes in 2026.
Common mistakes when building the worksheet
Most worksheets fail in the same handful of ways. None of the failures is dramatic, which is why the worksheet is so easy to get wrong by accident.
Using menu price instead of selling price
The price on the menu is rarely the price the kitchen actually receives. Happy hour discounts, loyalty discounts, online order commissions, and bundled prix fixe menus all change the effective revenue per plate. A worksheet that uses the printed price will overstate margin on any item that is regularly discounted. The cleaner input is the average selling price for the period, after discounts and before tax.
Forgetting the plate cost of the side
Many dishes are sold as an entrée with a side. If the side cost is not in the plate cost, the contribution margin is overstated. The fix is to add the average side cost, or to build the side into the recipe. Soups, salads, and bread baskets are the most common line items that quietly disappear from a plate cost build.
Counting only dinner service
A worksheet built from dinner data alone will misclassify lunch items and breakfast items, which often have very different contribution margins. If the menu is served all day, the worksheet should reflect all day. If the lunch menu is separate, it should have its own worksheet.
Moving the threshold lines after the fact
The thresholds are the whole point of the worksheet. If they are shifted to put a favorite dish in a flattering quadrant, the matrix becomes a justification rather than a diagnostic. The threshold should be set first, the dishes should be plotted second, and the conversation about changes should come third.
Confusing contribution margin with gross profit
Contribution margin does not include labor, rent, or utilities. It is the dollars available to cover those costs, not the dollars that remain after them. A star with a strong contribution margin is still a star in the worksheet, but a star with a strong contribution margin and a labor-heavy plate is a candidate for a separate labor review. Mixing the two is a common source of cross-talk between the kitchen and the finance team.
How the worksheet connects to menu design
The worksheet is a backend tool, but its output is meant to drive what the guest actually sees. Three connections are worth understanding.
Placement and the eye path
Studies of menu reading behavior consistently find that guests look at the top right of the first page first, the center of the first page second, and the top right of the second page third. Stars that are placed outside that eye path are losing easy money. The worksheet identifies the stars. Menu design places them in the eye path.
Description and price anchoring
A strong star can carry a higher price if the description is doing its job. The same dish with a one-line description and a three-line description often moves through a different range of contribution margins. The worksheet tells the operator which dishes are worth investing in, description-wise. A puzzle with a careful description rewrite can become a star without changing a single ingredient.
Photo, icon, and callout policy
Photos and callout boxes sell, but they sell every dish they touch, including the dogs. A blanket policy of marking every dish with a small icon will lift the dogs along with the stars. The worksheet gives the operator a way to be selective: mark the stars, leave the plowhorses alone, and use the callouts to point guests in the direction the matrix already recommends.
Adapting the worksheet for different restaurant types
The basic four-quadrant shape works across formats, but a few adjustments make it more useful in specific contexts.
Quick service and fast casual
In quick service, contribution margin per item is small, which means small percentage shifts have an outsized effect. A worksheet that uses units sold can be paired with a parallel worksheet that uses transactions, since many guests order a combination. Plowhorses in this format often point to combo or bundle opportunities, and the action is to assemble a star and a puzzle into a value combination rather than to reprice either dish on its own.
Full service and fine dining
Full service menus tend to be smaller and the contribution margins wider, so the thresholds sit further apart. In fine dining, a dish that sells only a few times per night can still be a star if the margin is strong, and the menu is small enough that one puzzle can distort the average. Median thresholds and a smaller set of comparison items help. A short tasting menu also works well with a per-cover variant of the worksheet that treats the tasting as a single item with internal cost layers.
Bars and beverage programs
Beverage menus work in the same shape, with pour cost in place of plate cost. Stars are the drinks that move quickly and that earn above-average pour margin. Plowhorses are the well liquors and the house wines that the bar moves in volume. Puzzles are the premium spirits and the small-production wines. The actions are similar, but the time horizon is shorter, since a beverage menu can be re-costed and reprinted much faster than a food menu.
Catering and event menus
Catering has its own quirks. Most revenue is committed in advance, and the same dish may appear on several different event menus at different price points. The worksheet should be built per price tier, since a star at a corporate lunch is not necessarily a star at a wedding. A small set of truly versatile dishes tends to emerge, and that set is worth tracking separately from the rest of the recipe library.
How often the worksheet should be revisited
For a stable menu, a quarterly worksheet is the right cadence. For a menu that is actively changing, monthly is better. For a brand new menu, a worksheet two to three weeks after launch gives the kitchen a clean read on whether the new items are settling, and a second one at six weeks confirms the pattern.
The worksheet should also be revisited after any structural change. A new point-of-sale system, a new ingredient supplier, a major price move on a key protein, or a shift in service style from table service to counter service all change the inputs enough to make an old worksheet misleading. A quick pass through the matrix after each change is faster and cheaper than a year of drifting decisions.
How the worksheet fits into a larger restaurant growth program
Menu engineering is one of the highest-leverage activities in a restaurant, because the menu is the only part of the operation that is sold to every guest, every time. Marketing can bring new guests in, but the menu decides how much each one leaves behind. The worksheet is the operating rhythm that keeps that decision honest.
It also pairs well with adjacent work. A menu description audit uses the same matrix to figure out which dishes deserve the heaviest writing investment. A restaurant marketing review uses the same star and plowhorse list to decide what to feature in ads and social posts. The worksheet is the input. The other programs are the way the operator turns the input into money.
Frequently asked questions
What is a restaurant menu engineering worksheet?
A restaurant menu engineering worksheet is a one-page working document, usually a spreadsheet, that lists every dish on a menu with its plate cost, contribution margin, and sales mix, then plots each dish on a four-quadrant matrix of profitability and popularity. It turns menu decisions from instinct into a structured review that any operator can run on a regular cadence.
How long should the data period for the worksheet be?
One to four weeks is the most common range. A single week captures both a busy and a quiet day, which makes it a reasonable sample for most restaurants. Multi-unit operators often use a four-week period to smooth out local events and weather swings. The data should always come from the same period as the recipe costs used to build plate food cost.
Do I need special software to fill in a menu engineering worksheet?
No. A basic spreadsheet with two inputs per dish is enough. The contribution margin column, the sales mix column, and the quadrant label are the only calculations the worksheet actually needs. Restaurants that already use a modern point-of-sale system with an item-level reporting view can usually pull the sales numbers in a few minutes, which makes the worksheet faster to update each cycle.
What is the difference between a star and a plowhorse?
A star is a menu item that earns above-average contribution margin and sells above its share of the menu. A plowhorse sells well but earns below-average contribution margin. Stars are the items the menu should protect and feature. Plowhorses are the items the menu should engineer through small portion, ingredient, or price adjustments, because the popularity is an asset the operator does not want to lose.
What does it mean if most of my menu falls into the dog quadrant?
It usually means the menu has not had a structured review in a long time. A menu that is mostly dogs is telling the operator that the mix is out of step with current guest demand or current ingredient costs. The first move is to identify a small number of candidates for removal and a small number of replacement candidates from the kitchen team’s pipeline, then re-run the worksheet after the changes settle.
How does the worksheet account for discounts and promotions?
The worksheet works best when the contribution margin is built from the average selling price for the period, not the printed menu price. If a dish is sold at a discount on certain days, the discounted revenue is the number that goes into the margin column. Treating the printed price as the selling price will quietly overstate margin on heavily discounted items.
Can the worksheet be used for beverages as well as food?
Yes. The same matrix shape works for cocktails, wine, beer, and non-alcoholic drinks, with pour cost in place of plate food cost. The thresholds are set from the beverage menu’s own averages, and the four quadrants have the same meaning. Bar and beverage programs often revisit the worksheet more frequently than food menus, since pour cost and supplier pricing can change quickly.
How does menu engineering 2.0 differ from the classic worksheet?
Menu engineering 2.0 keeps the same matrix but adds a guest satisfaction input, usually from comment cards, online reviews, or server feedback. The result is a more nuanced read of the menu in which a star that is dragging down satisfaction can be flagged even if it earns a strong margin. The classic worksheet is a financial tool. Menu engineering 2.0 is a financial and experience tool layered on top of it.
What should I do with a dish that sits on the boundary between two quadrants?
Borderline dishes are common, and the right answer is usually to leave them in the worksheet for a second cycle rather than to engineer them on the first pass. A dish that is right on the average contribution margin line may move clearly into star or plowhorse territory once ingredient costs shift or once the description is rewritten. Patience is part of the discipline.
How long does it take to build a worksheet from scratch?
For a small menu, the first worksheet usually takes a few hours, mostly because of the plate cost build. The recipe cost step is the slowest part. Once a recipe library exists, the second and third worksheets are much faster, since the only new work is pulling sales data and updating the few ingredients that have changed price.