Menu engineering: one 10-dish menu, worked through

Menu engineering sorts each dish by two numbers: how many plates you sell, and how many dollars each plate leaves after its food cost. Two lines split the menu into four groups: stars, plowhorses, puzzles and dogs. In the worked menu below, the best seller ties for the lowest food cost percentage on the list and earns the fewest dollars per plate of all ten dishes.

What menu engineering measures

The method comes from Michael L. Kasavana and Donald I. Smith. A 2007 review of menu analysis methods dates it to 1982, and a library record lists a revised 1990 edition of their book, Menu Engineering: A Practical Guide to Menu Analysis. We couldn't open the book itself, so every rule below is as three articles in a peer-reviewed hospitality journal, FIU Hospitality Review, describe it.

It asks two questions about each dish, compared with the other dishes in the same group (mains with mains):

• Is it popular? Its share of all the plates sold in the group.
• Does it earn its keep? Its contribution margin: the menu price minus the plate cost. It's the dollars one plate leaves to pay for wages, rent and profit.

Plate cost means the food on the plate. A 1995 article by three University of Central Florida hospitality faculty, LeBruto, Quain and Ashley, adds that any other cost that goes out with each plate, such as paper goods, belongs in it too. So if a dish leaves in a to-go box, count the box. They also sum up why the method looks at dollars: “operators deposit cash, not percentages”.

The two cut-off lines, as the method sets them

The popularity line. Take an equal share: 100% divided by the number of dishes in the group. A dish counts as popular when its share of plates sold is at or above 70% of that. LeBruto, Quain and Ashley work an eight-item example: 100% ÷ 8 × 70% is 8.75%, so a dish is popular at 8.75% of plates sold or more. A 2001 study by Horton puts it the same way: at, or above, 70% of the average.

The profit line. Work out each dish's contribution margin, multiply it by the plates sold, add those up, and divide by all the plates sold. That's the menu's weighted average contribution margin. A dish above it counts as high; below it, low. LeBruto, Quain and Ashley and a 2007 review by Taylor and Brown both describe it this way. Horton's own study used a plain average of the dishes' margins instead, which can draw the line in a different place. We use the weighted one.

One menu, worked all the way through

Here's Larkspur Kitchen, a made-up independent restaurant. Every name and number is an example. It has ten mains and sold 2,000 of them last month. The plate costs are from its recipe cards at this month's invoice prices.

The month comes to $43,050 in sales of mains and $13,050 in plate costs, a food cost of 30.3% ($13,050 ÷ $43,050). Take one from the other and you get the same $30,000 as the last column adds up to.

The shares are rounded, so they add to 100.1%. Look at the plowhorses: they're 900 of the 2,000 plates and more than a third of the margin. The spaghetti alone left $4,032 of margin, the third-largest total on the menu, behind only the roast chicken and the salmon. A dish can earn little per plate and still matter a lot in total, so a plowhorse is something to work on, not to drop.

Why the dollars decide, not the food cost percentage

On a food cost sheet, the spaghetti looks like the best dish on the menu: 20%, tied for the lowest, and the biggest seller. By dollars, it earns $11.20 a plate, the least of all ten. The ribeye has the highest food cost on the menu at 40%, and each plate leaves $22.80, more than twice as much.

The mushroom risotto is the clearest case. Its food cost is 20% too, and it still lands in the dog group: $14.40 a plate is under the $15.00 line, and 120 plates is under 140. A low percentage on a low price is a small number of dollars. LeBruto, Quain and Ashley make the same point with a $10 dish that costs $5 (50%, $5 left) against a $6 dish that costs $2 (33%, $4 left).

The food cost guide shows how to cost one plate. The percentage is still worth watching for the kitchen as a whole. It just can't rank dishes against each other on its own.

What to try with each group

LeBruto, Quain and Ashley list the actions Kasavana and Smith suggest for each group. Treat each one as a test: change one thing, wait a month, and run the numbers again. None of them is sure to work, and guests decide.

Stars (roast chicken, salmon, ribeye). The method suggests testing a higher price on them. Keep the recipe and the portion the same while you do. Watch how much room a star has: the ribeye sold 150 plates, only 10 above the 140 line. At $2 more it leaves $24.80 a plate, and it needs 138 plates to earn the month's $3,420 again. That's below the line, so it could slip into the puzzle group while earning the same.

Plowhorses (spaghetti, cheeseburger, fish tacos). The method suggests raising the price toward the line, as long as the dish stays popular. LeBruto, Quain and Ashley warn about dishes far below the line: a raise big enough to get them there can cost them their popularity. The spaghetti sits $3.80 under today's line, and because a raise on 360 plates lifts the average too, it would take about $4.65 more (a 33% raise) at the same sales to clear it. Try a smaller step, or bring the plate cost down. Taylor and Brown list retooling the recipe to cut its cost as the other option. That might be a cheaper side, or bringing the portion back to what the card says.

How many guests a raise can lose before it costs you is worked out in how much to raise menu prices.

Puzzles (short rib, pork chop). The method suggests featuring them on the menu or lowering the price. Taylor and Brown mention where they sit on the menu and how you promote them. Move the short rib to the top of the mains, or have servers mention it, for one month. Count the gain honestly. If 30 more guests order the short rib, that isn't 30 × $20.80 = $624 extra if they'd otherwise have had the spaghetti. In that case it's 30 × ($20.80 − $11.20) = $288.

Dogs (mushroom risotto, grain bowl). The method suggests taking them off or rotating them with similar dishes. Ask one question first: are they the only meat-free mains? If so, they may be why some tables come in, and the sort can't see that. The risotto is only $0.60 under the line, so $1 more would put it at $15.40 a plate, above the line. Re-cost it first; a dog with a wrong plate cost is just a mistake on paper.

What the method can't see

• One period's sales. A month is a snapshot. A summer salad and a winter braise won't look the same in July, so compare like months, or run it again each season.
• New dishes have no history. A dish added mid-month, or a special, hasn't had the same chance to sell. Leave it out or mark it until it has a full month.
• Labor and kitchen time. LeBruto, Quain and Ashley list this as a main weakness: the margin counts only food. When Horton timed the work on each dish in one restaurant, adding labor moved 6 of 52 dishes into another group. One salad took $1.16 of labor against $1.07 of food. Food and labor together are your prime cost (see the prime cost guide).
• The group you compare in. In Horton's study, looking at starters, salads and sandwiches, and entrees separately changed the group of 32 of the 52 dishes. Horton's point is to group dishes the way your guests order them: if a starter or a salad often stands in for a main, sort it with the mains.
• The lines move. Both lines come from your own menu's averages, so some dish always sits below the profit line, and moving one dish moves the lines. Change one dish and the others can change group too, as Horton and Taylor and Brown both note.
• Price isn't everything. Taylor and Brown warn that chasing margin can favor higher-priced dishes, and those can lower demand over time.
• Plate costs have to be right. A card costed at last year's prices, or a portion heavier than the card, moves a dish across the line on paper. See actual vs theoretical food cost for checking a card against what really leaves your stock.

This week

1. 01Pull last month's item report from your POS: how many of each main you sold. Mains only, and leave out dishes that weren't on for the whole month.
2. 02Get a current plate cost for each main, at this month's invoice prices. Add the to-go box for anything that leaves in one.
3. 03Work out each dish's margin (price minus plate cost) and its month (margin × sold). Add the month column and divide by the plates sold: that's your profit line.
4. 04Work out the popularity line: 100% ÷ the number of mains × 70%, then times the plates sold. Sort each dish into star, plowhorse, puzzle or dog.
5. 05Pick one test for one dish: a small raise on a plowhorse, a better spot for a puzzle, or a re-cost of a dog. Write the date down and run the same sort next month.