Restaurant menu engineering: optimize yourmenu with the popularity x margin matrix

Origin and principle of menu engineering

Menu engineering was born in 1982 at the Cornell University School of Hotel Administration. Michael Kasavana and Donald Smith published a foundational article in the Cornell Hotel and Restaurant Administration Quarterly proposing a 2x2 grid inspired by the BCG (Boston Consulting Group) matrix. Instead of segmenting a product portfolio along growth and market share, they transposed the logic to a restaurant context: popularity (customers love it) and contribution margin (the restaurant earns from it).

The model's strength lies in its simplicity: two axes, four quadrants, one clear action per quadrant. It forces the chef-owner to step out of intuition and pilot the menu with two measurable inputs: the sales mix (how often each dish sells) and the food cost (what each dish costs to produce). It is a contribution analysis approach, distinct from ABC analysis which only considers revenue without weighing margin.

Menu engineering is not a fad: forty years on, it remains the reference taught in hospitality schools and used by Accor, Compass, Sodexo, Aramark, and the vast majority of restaurant consultants. Its power comes from combining calculation rigor with the visual readability of the 2x2 matrix, which turns a subjective conversation about the menu into a fact-based debate around four quadrants.

The four categories: Stars, Plowhorses, Puzzles, Dogs

Stars (high popularity + high margin)

Stars are the heroes of your menu. Customers love them and they generate above-average margin. Action: protect these dishes, do not change the recipe or the price, place them inside the menu's golden triangle (top right, top left, center), and train servers to recommend them. A Star whose mix collapses overnight is a strong warning sign: ingredient quality, plating, kitchen training.

Plowhorses (high popularity + low margin)

Plowhorses drive volume but eat margin. They are often the historic signature dishes of a venue, sold at a price that has not kept up with input costs. Action: reformulate quietly (swap a side for a more profitable one, adjust garnish portions, optimize the protein gram weight), nudge the price up 3 to 5 percent, control the food cost line by line via the spec sheet. Never bluntly remove a Plowhorse: it is a draining volume that can be transformed into a Star.

Puzzles (low popularity + high margin)

Puzzles enjoy enviable margins but do not sell. Action: amplify visibility. Reposition them on the menu (golden triangle), rename them to sound more appetizing, add a photo, ask servers to suggest them first, simplify the description. A well-repositioned Puzzle can flip to Star in two to three months. If nothing moves after a quarter, the dish is probably miscalibrated (technique, atypical ingredient, price too high for the format).

Dogs (low popularity + low margin)

Dogs neither please nor pay. Action: remove or radically rethink. Before deletion, verify that the dish is not a niche anchor (vegetarian, gluten free, kids menu). If not, the Dog clutters the menu, complicates production, lengthens the shopping list, and distracts servers. A 12-dish menu without Dogs sells better than an 18-dish menu containing 5 Dogs.

Building the matrix step by step

Repeatable quarterly method:

Step 1. Extract the sales mix of the past 28 days by category from the POS. If you use a restaurant ERP connected to the POS, this calculation is automatic. Otherwise, export to CSV and clean cancellations, complimentary items, and entry errors.

Step 2. Pull the unit food cost of every dish from its spec sheet. If the spec sheet is outdated (supplier change, raw material spike), recalculate first: a stale food cost distorts the entire matrix. This is the most common mistake of chefs starting with menu engineering: an 18-month-old spec sheet that no longer reflects reality.

Step 3. Calculate average popularity and the 70 percent threshold per category. Spreadsheet: average(volumes) then threshold = average x 0.70. Any dish above the threshold sits in the «popular» zone.

Step 4. Calculate the weighted-average contribution margin per category: sumproduct(margin ; volume) / sum(volume). Any dish above this average sits in the «high margin» zone.

Step 5. Place each dish on the matrix by crossing the two axes: popular + high margin = Star, popular + low margin = Plowhorse, low popularity + high margin = Puzzle, low popularity + low margin = Dog.

Step 6. List actions per quadrant and set a horizon: quarterly review to measure the impact. A menu does not transform in one month; six to twelve weeks are needed for a repositioning or a price increase to yield reliable data.

Worked examples: starters, mains, desserts

Starters (extract, 4 lines)

Salmon tartare, price 12.50, food cost 3.80, margin 8.70, mix 24 percent: Star. Slow-poached egg with bacon, price 9.50, food cost 2.10, margin 7.40, mix 19 percent: Star. House foie gras, price 18, food cost 6.20, margin 11.80, mix 8 percent: Puzzle (high margin but undersold). Seasonal gazpacho, price 8.50, food cost 3.80, margin 4.70, mix 6 percent: Dog.

Mains (extract, 4 lines)

Signature burger, price 17, food cost 5.20, margin 11.80, mix 26 percent: Star. Ribeye steak with fries 200 g, price 24, food cost 11.40, margin 12.60, mix 18 percent: Plowhorse (high volume, margin pulled down by meat cost). Truffle risotto, price 26, food cost 7.80, margin 18.20, mix 7 percent: Puzzle. Grilled fish of the day, price 22, food cost 9.80, margin 12.20, mix 5 percent: Dog (low sales, low margin after losses).

Desserts (extract, 4 lines)

Homemade tiramisu, price 7.50, food cost 1.40, margin 6.10, mix 32 percent: Star. Coffee gourmand, price 8.50, food cost 2.80, margin 5.70, mix 28 percent: Plowhorse. Chocolate souffle, price 9.50, food cost 1.90, margin 7.60, mix 9 percent: Puzzle. Aged cheese plate, price 10, food cost 5.40, margin 4.60, mix 6 percent: Dog.

How not to misread the matrix. Common bias: judging a dish on percentage margin alone. A cheese plate at 46 percent gross margin can be a Dog if it does not sell. Always reason in absolute margin x volume, i.e. total contribution to bottom line.

Pricing psychology and price increases

Pricing psychology is the toolbox of techniques that influence price perception without changing actual value. Four core techniques in restaurants.

Charm pricing. 14.50 perceived lower than 15. The academic literature (Anderson & Simester, 2003) confirms a several-percent volume effect on prices ending in 9 or 5. Caveat: avoid charm pricing in fine dining, where round prices (24, 32, 48) signal quality.

Currency symbol removal. A Cornell study (2009) shows removing the currency symbol increases average spend by 8 percent. The customer reasons in number, not in currency.

Decoy effect. Place a deliberately expensive item (the decoy) that is not meant to sell, but makes the rest of the menu feel reasonable by contrast. A 68-EUR lobster on a menu where mains sit at 22-28 EUR pulls the average ticket up.

Anchoring. Price ordering on the menu matters: starting each category with the most expensive dish anchors perception. The mid-range price then feels like a deal.

To raise prices without losing customers: prefer 3 to 5 percent twice a year (seasonal menu changes) over 10 percent in one shot. Camouflage the increase by reformulating Plowhorses (side change, plating). Keep 2 to 3 anchor dishes at stable prices (lunch formula, historic signature) to preserve value perception. Communicate added value: ingredient origin, label, technique, local sourcing. A dish moving from 14 to 16 is accepted if the menu now says «Brittany IGP salmon» instead of plain «salmon».

Seasonality and menu rotation

Three rhythms coexist in a well-piloted menu. Permanent menu: light change four times a year, quarterly menu engineering review, 1 to 2 dishes adjusted per review to avoid disorienting regulars. Seasonal menu: full overhaul at season changes (March 1, June 1, September 1, December 1), 6 to 8 new dishes per cycle, Stars systematically retained. Daily special or chalkboard: daily, acts as an incubator for testing future permanent dishes. Measure the mix of a special over 4 to 6 services before promoting it to the menu.

Raw-material seasonality directly impacts food cost and therefore the matrix. A winter tomato costs 2 to 3 times more than a summer tomato and tastes worse: never freeze a tomato recipe on a permanent menu. Seasonal rotation is not a chef's whim; it stabilizes food cost at a low level and prevents the menu from drifting into a string of Plowhorses as input prices rise.

Link to restaurant production: menu changes must align with production capacity, team rota, and remaining stock. A March 1 change that pulls beef bourguignon must be communicated 3 weeks ahead to the production team to avoid meat overstock.