Metrics, Costs and Finance

Menu Engineering: Stars, Plowhorses, Puzzles, Dogs

Sort every menu item by popularity and contribution margin, then decide what to promote, reprice, rework or cut, with a worked eight-item Example.

AssistCall research teamPublished Reviewed by Faruque Amin, VP of Sales and Marketing7 min read

On this page
  1. What menu engineering tells you
  2. The data you need before you start
  3. Build the matrix in six steps
  4. Worked Example: an eight-entree dinner menu
  5. What to do with each quadrant
  6. Example: turning the grid into decisions
  7. Add labor and prep before you act
  8. Why beef-heavy plates need a closer look now
  9. Menu design moves that shift the mix
  10. Mistakes that skew the results
  11. How often to rerun it
  12. Frequently asked questions
  13. References
Metrics, Costs and Finance: Menu Engineering: Stars, Plowhorses, Puzzles, DogsArticle coverMETRICS, COSTS AND FINANCEMenu Engineering:Stars, Plowhorses,Puzzles, DogsAva for RestaurantsInsightsMetrics, Costs and Finance: Menu Engineering: Stars, Plowhorses, Puzzles, DogsArticle coverMETRICS, COSTS AND FINANCEMenu Engineering:Stars, Plowhorses,Puzzles, DogsAva for Restaurants

Key takeaways

  • Menu engineering ranks each item on two measures at once: how often it sells and how many dollars it leaves after food cost.
  • Food and labor each take roughly 33 cents of every sales dollar, so the mix on your menu decides much of what is left[1].
  • Judge items by contribution margin in dollars, not food cost percentage; a high food cost item can still be your best earner.
  • Rerun the matrix whenever supplier prices move sharply; beef and veal prices are forecast to rise 8.7% in 2026[2].

What menu engineering tells you

Most owners know their best sellers and can guess their most profitable plates. Menu engineering puts both facts on one grid so you can see which items carry the menu, which are popular but thin, which earn well but hide, and which do neither. The output is a short list of decisions for the next menu print: what to feature, what to reprice, what to rework and what to remove.

The stakes are real. Food and labor are a restaurant's two largest costs, each taking roughly 33 cents of every sales dollar[1], and 42% of operators said their restaurant was not profitable in 2025[3]. A menu that sells more of the right plates is one of the few levers that improves margin without adding a single guest.

The data you need before you start

  • Item sales counts for one menu period, ideally four weeks, pulled from the POS by item rather than by category.
  • Current menu price for each item, net of any built-in discounts such as happy hour pricing.
  • A current plate cost for each item from an up-to-date recipe card, including garnish, sides and sauces.
  • A note of anything unusual in the period: a holiday, a closure, an item that was 86'd for days.

Use one menu section at a time. Comparing a side salad with a ribeye says little; comparing entrees with entrees tells you where guests spend and what each choice is worth.

Build the matrix in six steps

  1. Calculate contribution margin

    For each item, subtract plate cost from menu price. This is the dollar amount the item contributes toward labor, rent and profit each time it sells.

  2. Calculate menu mix

    Divide each item's count by the total count for the section. This share is the item's popularity.

  3. Set the popularity line

    The common convention marks an item as popular if its mix share is at least 70% of an equal share. With eight items an equal share is 12.5%, so the line sits near 8.75% in the Example below.

  4. Set the margin line

    Multiply each item's margin by its count, add the results, and divide by the total count. That weighted average margin is your line between high and low earners.

  5. Classify every item

    Stars sit above both lines, plowhorses are popular with low margin, puzzles earn well but sell little, and dogs are below both lines.

  6. Write one decision per item

    Keep, feature, reprice, re-portion, rename, move or remove. If you cannot name the action, the analysis is not finished.

Worked Example: an eight-entree dinner menu

This is a fictional Example for a full-service restaurant over four weeks. Total entree count was 2,050 and total contribution was $29,625, so the weighted average contribution margin is about $14.45 and the popularity line is about 8.75% of mix.

Example: menu engineering results for a fictional eight-entree dinner menu over four weeks
ItemSoldMixPricePlate costMarginClass
Steak frites21010.2%$34.00$14.20$19.80Star
Roast chicken38018.5%$24.00$6.30$17.70Star
Salmon bowl24011.7%$26.00$9.40$16.60Star
Smash burger62030.2%$16.00$5.10$10.90Plowhorse
Fish tacos30014.6%$18.00$6.90$11.10Plowhorse
Braised short rib904.4%$32.00$11.50$20.50Puzzle
Mushroom risotto1105.4%$21.00$4.60$16.40Puzzle
Cobb salad1004.9%$17.00$6.80$10.20Dog

Notice the steak in this Example. Its food cost is about 42% of price, the worst on the menu, yet it is a star because it leaves $19.80 every time it sells. The risotto has the lowest food cost at about 22% and still sits in the puzzle box because too few guests order it. If you managed this menu on food cost percentage alone, you would push the wrong plates.

What to do with each quadrant

Typical moves for each menu engineering class
ClassWhat it meansMoves to considerWatch out for
StarPopular and earns above averageKeep quality and portion steady, give it the best spot on the menu, train servers to describe itQuiet portion creep or a cheaper substitute that guests notice
PlowhorseGuests love it but it earns littleSmall price step, re-plate with a lower-cost side, offer a paid upgrade, pair it with a high-margin add-onRaising price so far that loyal guests stop ordering it
PuzzleEarns well but few people order itMove it to a prime position, rename or rewrite the description, add a photo online, let servers taste itKeeping a puzzle with costly prep that also creates waste
DogLow sales and low marginRemove it, or rework the recipe and price, or keep only if it serves a real purpose such as a dietary optionCutting the only vegetarian or gluten-free choice on the section

Example: turning the grid into decisions

Classification is not the finish line. Each item in the fictional Example gets one written decision that the chef and manager agree on before the next print:

  • Steak frites, roast chicken and salmon bowl (stars): no price change this round, lock portions with a photo spec, and move the chicken to the top of the entree list where it is seen first.
  • Smash burger (plowhorse): a modest price step plus a paid upgrade, such as a third patty or a premium cheese, so heavy users can choose to spend more.
  • Fish tacos (plowhorse): swap the side to a lower-cost slaw that tests well and keep the price, which protects a popular lunch choice.
  • Braised short rib (puzzle): rename it, rewrite the description around the long braise, add a photo to the online menu and have servers mention it on weekends.
  • Mushroom risotto (puzzle): keep it as the vegetarian anchor, but place it higher and pair it with a wine suggestion.
  • Cobb salad (dog): rework into a seasonal salad with the same base ingredients, or remove it and fold its produce into other dishes.

Write the decisions in a one-page brief with the reason for each and the date you will check the result. When the next four weeks of data arrive, compare each item's count and margin with the brief so you learn which kind of move works for your guests.

Add labor and prep before you act

The classic matrix uses food cost only, which can mislead in a scratch kitchen. A puzzle that needs a two-day braise ties up oven space and a skilled cook, and a star that is plated in a minute is better than its margin alone suggests. Before you promote a puzzle, ask three questions. How many minutes of prep and line time does each plate need? Does the dish share ingredients with other items, or does it need its own inventory that spoils if it does not sell? Can the line produce it consistently at your busiest hour? If any answer is poor, fix the recipe or the process before you push sales toward it, or you will trade a margin gain for slower tickets and more waste.

Why beef-heavy plates need a closer look now

Menu engineering is only as good as the plate costs behind it, and those costs are moving unevenly. USDA forecasts beef and veal prices to rise 8.7% in 2026, well ahead of menu prices for food away from home and grocery prices overall[2].

USDA 2026 price forecasts: beef versus restaurant and grocery pricesForecast percent change in 2026
USDA 2026 price forecasts: beef versus restaurant and grocery pricesBar chart. Beef and veal: 8.7%; Food away from home: 3.5%; Food at home: 2.3%.Beef and vealBeef and veal: 8.7%8.7%Food away from homeFood away from home: 3.5%3.5%Food at homeFood at home: 2.3%2.3%USDA 2026 price forecasts: beef versus restaurant and grocery pricesBar chart. Beef and veal: 8.7%; Food away from home: 3.5%; Food at home: 2.3%.Beef and vealBeef and veal: 8.7%8.7%Food away from homeFood away from home: 3.5%3.5%Food at homeFood at home: 2.3%2.3%

Source: USDA Economic Research Service, 2026[2]

Show the data
ItemValue
Beef and veal8.7%
Food away from home3.5%
Food at home2.3%

Restaurant menu prices rose 3.5% in the 12 months to September 2026[4], so a burger or steak whose plate cost rises faster than its price slides down the margin axis without anyone touching the menu. Since February 2020, restaurant hourly earnings have risen 41% and wholesale food prices 35%[5]; a matrix built on last year's recipe cards will flatter exactly the items that most need attention. Update the costs first, then classify.

  • Put stars and the strongest puzzles where eyes land first: the top of a section, a boxed feature, or the first item a server mentions.
  • Drop the dollar sign and the trailing zeros so prices read as quieter information rather than the headline.
  • Avoid lining prices up in a right-hand column, which invites guests to scan for the cheapest plate.
  • Write descriptions for puzzles that explain what makes the dish worth ordering; keep plowhorse descriptions short.
  • Mirror the changes on your online menu and third-party listings, which many guests read before they arrive.

Mistakes that skew the results

  • Mixing menu sections, so a high-volume appetizer drags the popularity line for entrees.
  • Using stale plate costs, especially for proteins, dairy and produce that reprice weekly.
  • Ignoring labor: a puzzle that needs fifteen minutes of skilled prep may earn less than its margin suggests.
  • Reading one slow week as a trend. Use at least four weeks and exclude days when an item was unavailable.
  • Removing a dog that drives party decisions, such as the one dish a vegan guest can order.

How often to rerun it

Run the full matrix before every menu reprint and a quick version every quarter. Between those, watch three triggers: a supplier price jump on a key protein, a sustained drop in a star's count, and any change to portion or garnish. Each one can move an item across a line. Pair the results with your plate cost cards and your pricing plan so that every change reaches the printed and online menus at the same time.

Frequently asked questions

What are stars, plowhorses, puzzles and dogs?

They are the four boxes of the menu engineering matrix. Stars sell well and earn above-average contribution margin, plowhorses sell well but earn less, puzzles earn well but sell little, and dogs are low on both.

Should I use food cost percentage or contribution margin?

Use contribution margin in dollars to rank items. Food cost percentage is still useful for spotting recipe problems, but a high-priced plate with a high food cost percentage can leave more dollars per sale than a cheap plate with a low one.

How many weeks of sales data do I need?

Four weeks is a practical minimum for most menus. Exclude days when an item was unavailable and avoid periods dominated by a holiday menu.

Do I have to remove every dog?

No. Rework or reprice it first, and keep it if it serves a clear purpose such as a dietary option that lets a group choose your restaurant.

References

  1. National Restaurant Association (2026). Restaurants remain resilient despite challenging business conditions (opens in a new tab) ↑

  2. USDA Economic Research Service (2026). Food Price Outlook - Summary Findings (opens in a new tab) ↑

  3. National Restaurant Association (2025). Persistent Cost Increases and Enduring Demand Will Shape the Restaurant Industry in 2026 (opens in a new tab) ↑

  4. US Bureau of Labor Statistics (2026). Consumer Price Index: series CUUR0000SEFV (opens in a new tab) ↑

  5. National Restaurant Association (2026). Restaurants remain resilient despite challenging business conditions (opens in a new tab) ↑

Book a demo

See Ava handle your calls.

Book a short demo. We’ll walk through your calls and workflow, and agree what Ava should take care of.

Book a demo