Menu engineering for cocktail bars
Menu engineering sorts each drink on your list by how many it sells and how much it earns after cost. It shows which drinks to keep, price higher, move on the menu or think about cutting.
If each drink on your list has a costed A drink's exact recipe: each ingredient and how much of it., you know what it costs. Your Point-of-sale system: the register or software that records every sale. already counts how many of each drink sold. With those two numbers, menu engineering can show that a drink with a low The cost of a drink's ingredients divided by its menu price, as a percentage. earns less than you think.
What is menu engineering?
Menu engineering judges each item on a menu by two numbers: how many sold, and its contribution margin. Michael Kasavana and Donald Smith of Michigan State University published the method in 1982.
Take a Negroni that sells for $15 and costs $2.40 to make. Its contribution margin is $15.00 − $2.40 = $12.60: the money left from the price after you pay for what goes into it. Many bars call this number the The menu price of a drink minus what its ingredients cost. per drink.
Contribution margin = menu price (before tax) − cost of the drink
Contribution margin is not the same as pour cost. Pour cost is a percentage. Contribution margin is in dollars.
The four categories of the menu engineering matrix
Menu engineering draws two lines across your list: a popularity line, called the hurdle, and an earnings line, the average contribution margin. Every drink sits above or below each line, which gives four groups.
These categories compare drinks on your own menu. High and low don't mean profitable or unprofitable in absolute terms. They mean above or below the dividing lines calculated from the drinks you're comparing.
| Contribution margin at or above average | Contribution margin below average | |
|---|---|---|
| Sells at or above the 70% hurdle | Star | Plowhorse |
| Sells below the 70% hurdle | Puzzle | Dog |
Start with the hurdle. A drink's share of all the drinks sold is called its mix. Say your list has 10 drinks. If all 10 sold the same number, each would have a mix of 10% (100 ÷ 10). The hurdle is 70% of that, which is 7%, so any drink with a mix of 7% or more counts as popular. Kasavana and Smith chose 70%, not 100%, so a drink with a mix of 8% or 9% still counts as popular.
Mix % = number of the drink sold ÷ number of all drinks in the category sold × 100
Popularity hurdle = (100% ÷ number of drinks) × 0.70
The second line is the weighted average contribution margin. Each drink's total contribution is its contribution margin times the number sold. Add these up for the list, then divide by the total number of drinks sold. So a drink that sells more counts for more. Any drink at or above this average counts as high-earning.
Weighted average contribution margin = sum of (contribution margin × number sold) ÷ total number sold
How to run menu engineering on your list
Here are the six steps, on a list of 10 cocktails.
| Step | What you do | The example |
|---|---|---|
| 1 | Pick one category and a period of at least four weeks | Cocktails only, the last 28 nights |
| 2 | Take the number sold of each drink from your POS | 4,000 cocktails across 10 drinks |
| 3 | Work out the cost of each spec and subtract it from the price before tax | Negroni: $15.00 − $2.40 = $12.60 |
| 4 | Work out each drink's mix and the popularity hurdle | Hurdle 7%, 280 drinks |
| 5 | Work out the weighted average contribution margin | $11.29 |
| 6 | Sort the drinks into the four groups and choose one action for each | 4 stars, 3 plowhorses, 1 puzzle, 2 dogs |
And these are the 10 cocktails over the last 28 nights.
| Drink | Sold | Mix (%) | Price ($) | Cost ($) | Contribution margin ($) | Total contribution ($) | Group |
|---|---|---|---|---|---|---|---|
| Espresso Martini | 700 | 17.5 | 14.00 | 2.50 | 11.50 | 8,050 | Star |
| Negroni | 600 | 15.0 | 15.00 | 2.40 | 12.60 | 7,560 | Star |
| Daiquiri | 600 | 15.0 | 12.00 | 1.76 | 10.24 | 6,144 | Plowhorse |
| Aperol Spritz | 500 | 12.5 | 12.00 | 3.47 | 8.53 | 4,265 | Plowhorse |
| Old Fashioned | 400 | 10.0 | 15.00 | 2.42 | 12.58 | 5,032 | Star |
| Paper Plane | 400 | 10.0 | 14.00 | 3.42 | 10.58 | 4,232 | Plowhorse |
| Freezer Martini | 300 | 7.5 | 16.00 | 2.20 | 13.80 | 4,140 | Star |
| Last Word | 200 | 5.0 | 16.00 | 3.78 | 12.22 | 2,444 | Puzzle |
| Bijou | 150 | 3.75 | 15.00 | 3.90 | 11.10 | 1,665 | Dog |
| Pisco Sour | 150 | 3.75 | 14.00 | 3.03 | 10.97 | 1,645.50 | Dog |
| All cocktails | 4,000 | 100 | 11.29 | 45,177.50 |
Together they sold 4,000 and brought in $45,177.50 of contribution. So a cocktail needs 280 sold (7% of 4,000) to count as popular. The weighted average is $45,177.50 ÷ 4,000 = $11.29 a drink.
Two results may surprise you. The Daiquiri's pour cost is low: $1.76 ÷ $12.00 = 14.7%. It sold enough to clear the 280 hurdle, so it is popular. But at $12 it earns $10.24, which is $1.05 under the average, so it is a plowhorse. The Freezer Martini sold only 300, yet it is a star. Poured from a pre-batch, it costs $2.20 and sells for $16, so it earns $13.80, the most on the list. And its 300 sold still clear the 280 hurdle.
What to do with stars, plowhorses, puzzles and dogs
Kasavana gives a standard action for each group. Treat it as a starting point to check with your team, not an automatic decision.
| Group | Where to start | In the example |
|---|---|---|
| Star | Keep it prominent and keep it consistent. Check whether its price and demand still leave room for improvement | Espresso Martini, Negroni, Old Fashioned, Freezer Martini |
| Plowhorse | Test a modest price increase, or look for ways to reduce its ingredient cost without compromising the drink | Daiquiri, Aperol Spritz, Paper Plane |
| Puzzle | Move it on the menu, and ask the team to recommend it | Last Word |
| Dog | Find out who orders it and why. Keep, change or remove it based on what it contributes to the overall menu | Bijou, Pisco Sour |
Plowhorses: test a small price rise
Guests already order a plowhorse, so a small price rise earns you more on every one you sell. But if the price goes up too much, sales can fall below the hurdle and the drink becomes a dog. Here are price rises on the Paper Plane and the Aperol Spritz.
| Drink | Price now ($) | Price after ($) | Contribution margin after ($) | Extra over 28 nights ($) |
|---|---|---|---|---|
| Paper Plane | 14.00 | 14.50 | 11.08 | 200 |
| Aperol Spritz | 12.00 | 13.00 | 9.53 | 500 |
Together the two rises bring in $700 every 28 nights, about $9,100 a year. The Paper Plane now earns $11.08 and the Aperol Spritz $9.53. Both still earn under the list's average, so both stay plowhorses. That is fine: the aim is more dollars from drinks guests already want.
That $700 is the upside if guests keep ordering the same number. Check actual sales after the change. A drop in sales may reduce or erase the gain, while guests switching to other profitable drinks may offset some of the loss.
Puzzles: move it before you change the price
The Last Word earns $12.22 a drink, but it sold 200, under the 280 hurdle. So before you cut it or lower its price, try moving it. If the Last Word is buried in the middle of the list, move it to the beginning or end of its section. A café study in Tel Aviv found that items placed at either end of a menu section got about 20% more orders than when they sat in the middle. That makes placement worth testing, although the effect will depend on your menu and guests.
Then ask your team to offer the Last Word to guests who like the Negroni but want something sharper.
Dogs: ask who orders them
The Bijou and the Pisco Sour are the dogs. Both earn under the $11.29 average: the Bijou $11.10 a drink and the Pisco Sour $10.97. Together they sold 300 and brought in $3,310.50 of contribution.
If their guests switch to a star or the Last Word, the bar earns more per drink. But if those guests leave, the bar loses up to $3,310.50 every 28 nights. So before a dog goes, ask your team who orders it. Say a group of regulars comes in every Thursday for the Pisco Sour. For that bar, the drink earns more than its 150 sales show.
What the matrix can't tell you
One thing the matrix leaves out is labor. A costed spec counts only what goes in the glass. The Freezer Martini is poured from the freezer in seconds. The Pisco Sour needs an egg separated and a dry shake. Kasavana says an item's cost should include the labor to make it. So the Pisco Sour costs you more than its spec shows.
The matrix also doesn't show which drinks slow service down. A cocktail with a strong contribution margin may be less attractive on a busy night if it takes several minutes to make and holds up the station.
The matrix works on one kind of drink at a time: it can't fairly compare a cocktail with a beer. To see why, add beer to the cocktail list. A house lager sells for $7.00 and costs $1.05, so its contribution margin is $5.95. It sold 1,000 in the same 28 nights.
- The list now has 11 items and 5,000 sold (4,000 + 1,000).
- Total contribution is $51,127.50 ($45,177.50 + $5,950), so the average falls to $51,127.50 ÷ 5,000 = $10.23.
- The hurdle falls to (100 ÷ 11) × 0.70 = 6.36%, which is 318 drinks.
Here the drinks keep their prices from the 28-night table, before any price rise. Two plowhorses become stars: the Daiquiri ($10.24) and the Paper Plane ($10.58). Three cocktails become puzzles. The Freezer Martini, with 300 sold (6.0%), drops under the hurdle. The two dogs, the Bijou ($11.10) and the Pisco Sour ($10.97), now earn above the average. So five cocktails change group, though no guest ordered anything different. Keep cocktails, beer and wine in separate matrices.
Which nights you look at matters too. A period of 28 nights is enough to spot a drink that isn't selling. But before you cut one, look at the whole time the menu has been in use, so two slow weeks or a seasonal dip don't decide it.
Even over the menu's whole run, sales data can't tell you why a drink sold. Your team on the floor can: they know which drinks guests ask about, and which ones guests order again.
Free menu engineering worksheet
The worksheet is a costed menu in a spreadsheet. For each drink, you type the price, the cost and the number sold. It works out each drink's contribution margin, pour cost, mix and group, plus the popularity hurdle and the weighted average. A test-price column lets you try a new price and see the extra dollars at the same number sold.
It also counts your list by base spirit and by style, so you can see if too many drinks use the same spirit before you write the next menu.
The first sheet holds a worked example. The second is blank, with room for 30 drinks. Copy it once for cocktails, once for beer and once for wine.
The costs come from your specs. If you don't know a drink's cost yet, use the pour cost calculator to work it out.
For other spreadsheet apps, a plain CSV copy holds the worked example.
Questions
How often should I run menu engineering?
Run it once per menu, shortly before you change the menu, so the numbers can shape the next one. Between menus, look at your top sellers once a month.
Is menu engineering only for restaurants?
No. Kasavana and Smith built it on restaurant menus, but it needs only three numbers for each item: the number sold, the price and the cost. A bar with costed specs has all three.
Why does some software sort the same list differently?
Some tools, Jery among them, split the list at the median instead of the 70% hurdle and the weighted average. The median is the middle value, so about half the list sits on each side of each line. On the 10 cocktails in this article, the median is 400 sold and $11.30 contribution margin. Only the Freezer Martini moves: its 300 sold is under 400, so it goes from star to puzzle. When you compare two periods, use the same method and the same drink categories. Otherwise a change in group may come from the method, not from how the drinks sold.
Sources
- Kasavana, M. L. and Smith, D. I. Menu Engineering: A Practical Guide to Menu Analysis. Hospitality Publications, 1982. The original method; the book itself was not opened. The rules quoted here are from Kasavana's own two-part summary below.
- Kasavana, M. L., The Power of Menu Engineering, Part One (8 Apr 2025). The cost-percentage quote opens Part One; item cost including incremental labor is in the same article.
- Kasavana, M. L., The Power of Menu Engineering, Part Two: ME Analysis (14 May 2025). The 1 / N (70%) popularity rule, the average contribution margin rule and the four standard moves.
- Ivancsóné Horváth, Z., Kőmíves, C., Nagy-Keglovich, J. and Happ, É., Examining a menu on the basis of the Kasavana-Smith model in a Hungarian restaurant, Deturope 14(1), 2022, pp. 111 to 127. Miller (1980), LeBruto, Quain and Ashley (1995) and Taylor and Brown (2007) are cited from this review's literature section (pp. 113 to 114); those three works were not opened.
- Dayan, E. and Bar-Hillel, M., Nudge to nobesity II: Menu positions influence food orders, Judgment and Decision Making 6(4), 2011, pp. 333 to 342. Study 2, the café field study: 55% of an item's orders at the ends of its section against 45% in the middle (p. 339).
About the authors
Sarah co-founded and ran Teresa Cocktail Bar, a Tales of the Cocktail nominee. She was also the first customer success hire at Loaded and now builds Overproof.