The Ultimate Guide to Wine List Engineering

Written by Marco Azzari on 14 September 2026

Your best-selling wine is probably not your most profitable one, and your most profitable one is probably sitting in the cellar. Most lists I read have both problems at the same time, and nobody has noticed, because nobody has ever put the two numbers side by side.

Wine list engineering does exactly that. It's menu engineering, the method kitchens use to cost and rank every dish, applied to the wine list. By the end of this guide you'll have your own list plotted on one chart, every wine sorted into one of four boxes, and a clear next step for each one.

What is wine list engineering?

Wine list engineering is menu engineering for the wine list. You take every wine you sell, work out how much money each one leaves you per bottle or per glass, count how many you sold, and plot the two against each other. Every wine lands in one of four boxes, and each box comes with its own instruction: protect it, reprice it, push it, or cut it.

Food menus get this treatment all the time. A chef costs a plate down to the garnish, and a good general manager knows which entrées carry the kitchen. Wine rarely gets the same care. In a lot of restaurants I've worked in and around, the list was priced with one multiplier and a good feeling on the day it was printed, and nobody has read it line by line since.

The method asks two questions of every wine: how much does it earn you each time you sell it, and how often do you sell it? Neither number means much on its own. A bottle that earns $80 and sells forty times a year brings in less than one that earns $40 and sells four hundred times. You only see what the list is really doing when the two numbers sit on the same chart.

Why it's worth an afternoon of your time

Restaurants run on thin margins. Commonly cited industry benchmarks put a typical full-service restaurant at roughly three to five cents of profit per dollar of sales, and in a restaurant with a real wine program, wine is often one of the biggest lines on the beverage side.

At three cents on the dollar, small amounts of money carry a lot of weight. Take a restaurant doing $2 million a year in sales. At a 3% profit margin it keeps around $60,000, and say its wine sales are $200,000. Recover three points of margin on that wine and you've added about $6,000, which is a tenth of the year's profit.

Wine is also the cheapest place in the building to find that money. A plate passes through four or five pairs of hands before it reaches the table, while a bottle that's already in the cellar only needs opening. When you correct a price or a pour, nobody's shift gets longer, so whatever you recover goes straight to the bottom line.

The matrix is just as useful for what it tells you to leave alone. When a manager suspects the list is underperforming, the temptation is to change everything at once. The chart points you to the handful of wines that matter and lets you leave the rest alone.

And there's a benefit your guests will notice. A list that has been through this exercise is usually shorter and easier to read, because the wines nobody orders are gone and the good ones nobody understood finally get a proper description. Guests choose with more confidence, and your servers have a list they can actually recommend from.

Number one: the wine margin, in dollars

The first number is what each wine leaves you every time you sell it. The formula is simple:

Wine margin = menu price − bottle cost

Each part has a trap, so let's take them one at a time.

Start from the menu price, not the check total

US menu prices are listed before sales tax. The tax is added on top at checkout and passed on to your state and city, and the tip goes to your staff. Neither one is your revenue, so neither one touches your wine margin: you work straight from the price printed on the list. Rates vary by state, county and city, and some places tax alcohol at a different rate, so check your own with your state's revenue department or your accountant.

The one mistake to avoid here is pulling numbers from the wrong line of the POS report. Use net sales per item, not gross receipts that include tax, tips or service charges, or every wine will look better than it is.

Subtract what the bottle really cost

Use the price on your latest distributor invoice, net of any case or quantity discounts you actually received. In the three-tier system you buy from a distributor, and distributor prices move more often than people expect: the cost you remember from the day you listed the wine may be two price increases old.

For a bottle listed at $58 that cost $18: 58.00 - 18.00 = $40.00 of margin per bottle.

For glasses, divide by the glasses you really pour

By-the-glass needs one more step. Divide the bottle cost by the number of glasses you actually get out of a bottle, then subtract that from the glass price.

The number on paper is easy: 750 mL divided by your pour. Costing sheets usually assume five glasses of 5 oz. Real restaurants get fewer, often four, once you count the generous pours and the bottle opened for one table on Tuesday and poured down the drain on Thursday.

Here's the same Chardonnay sold at $18 a glass, on the old rule of thumb that one glass covers the wholesale cost of the bottle. At five glasses a bottle, each glass costs $3.60 and leaves $14.40. At four glasses, each costs $4.50 and leaves $13.50. That's 90 cents per glass you thought you were earning and never saw.

To find your real yield, spend two weeks comparing the glasses rung into the POS with the bottles opened for the glass program. Use that figure in the sheet, not the one on the costing card.

Why dollars, and not percentages

Most lists are judged by margin percentage, or its mirror image, pour cost, and percentages flatter cheap wines. In the example later in this guide, a Rioja earns 74% margin and a Barolo earns 60%, the lowest on the list. Yet every bottle of Barolo leaves $90 and every bottle of Rioja leaves $46.

Rent and payroll get paid in dollars. Percentages are useful for setting prices, but when you're deciding which wines deserve their place, the matrix has to run on the money each sale actually leaves behind.

Number two: how many you sold, and over which period

The second number is units sold per wine: bottles for the bottle list, glasses for the glass list. Your POS sales-by-item report has it. The harder question is which period to measure.

A full year is the best default. It captures seasonality on its own terms, from rosé and crisp whites on the patio in July to sparkling in December. Any shorter window rewards whatever happened to be in season and punishes the rest.

A season works if your restaurant or your list is seasonal. A ski-town restaurant open for the winter months, a beach-town patio, or a restaurant that prints a summer list and a winter list should compare each season with itself: this winter against last winter, not winter against summer.

A month is for a first read. If the list is new and there's no year behind it yet, or you've just made a change, a month gives you an early picture. Read it knowing the season is shaping the numbers.

A week is too short for the whole list. One table ordering six bottles of the same red can move it from one box to another. Use a week only to check a single change, such as a new pour size or a new wine by the glass, and compare it with the same week before the change.

Whichever period you choose, a few rules keep the comparison fair:

• Use the same period for every wine. Mixing a year for some lines and three months for others makes the chart meaningless.
• Scale up wines that weren't there the whole time. A wine listed halfway through the year, or out of stock for six weeks, lost those sales because it was missing from the list. Divide its sales by the weeks it was available and multiply by the weeks in the period, or leave it out until it has a full period behind it.
• Take out one-off events. A wedding or private dinner that ordered sixty bottles of the same white will turn a wine guests never choose into a Star.
• Leave comps and staff drinks out. They aren't sales. If your POS records them as zero-price items or voids, filter them before you copy the numbers. A comp is a decision, and it belongs on its own line.

Glasses or bottles: run two matrices

Run wine list engineering twice: once for the wines you sell by the glass, and once for the wines you sell by the bottle. Even when it's the same wine from the same case, a glass and a bottle are two different products.

The guest decides which one they want before they decide which wine. Someone ordering a glass is drinking alone, or wants something different with each course, or doesn't want to commit to a whole 750 mL bottle. A table ordering a bottle is choosing one wine to share for the whole meal. The glass list competes with the other glasses, and the bottle list competes with the other bottles. A guest who wants a glass of Chardonnay isn't weighing it against a bottle of Barolo.

The numbers behave differently too. A glass earns ten or fifteen dollars and a bottle earns forty or more. A glass carries the cost of whatever gets poured away when an open bottle goes off, and a bottle carries none. If you put both on one chart, every glass lands in the low-margin half, every bottle lands in the high-margin half, and the two lines that split the matrix end up somewhere that describes neither.

So a wine sold both ways gets two rows, one on each matrix, each with its own price, cost per serving, units sold and box. Don't be surprised when the boxes disagree. The Chardonnay can be a Star by the glass, because it's the obvious first order at the bar and the pour is priced well, and a Plowhorse by the bottle at the same time. That's two separate decisions: keep the glass as it is, and work on the bottle price.

Before you open Excel: three documents

You need three things on the desk before you start, and gathering them is usually the slowest part of the job.

1. The wine list as guests see it, with the current prices.
2. Your latest distributor invoices, so you have today's cost per bottle.
3. The POS report for your chosen period, showing units sold per item. Most POS systems export it to Excel or CSV. If you can't find the export, your POS provider or your accountant usually can in a few minutes.

Download the Excel template

The file does all the math for you: the margin per wine, the two lines and the box each wine falls into. There's a tab for bottles and a tab for glasses, and the matrix chart draws itself. It opens with a ten-wine example, so you can see how it works before you type over it with your own list.

Download the Excel template

Draw the two lines that split the list

The four boxes come from two lines: one for popularity and one for margin.

The popularity line is 70% of the average units sold. If every wine on the list sold equally, each one would sell the average. Menu engineering calls a wine popular when it reaches at least 70% of that fair share. The 70% is there because sales on any list are lopsided: a handful of wines do most of the volume, and with the plain average as the line, almost everything else would count as unpopular.

The margin line is the average margin per sale. Total margin divided by total units. It's weighted by sales on purpose, because it should tell you what a typical sale earns you, not what a typical line on the price list says.

In the example below, the ten wines sold 1,540 bottles in a year, an average of 154 each, so the popularity line sits at 107.8 bottles. Together they left $73,380 of margin, so the margin line sits at $47.65 per bottle.

A worked example: ten bottles, one year

Here's an illustrative bottle list for a mid-sized Chicago restaurant, measured over twelve months. Menu prices are before sales tax, costs are what the distributor invoiced.

Wine	Menu price	Cost	Sold	Margin per bottle	Total margin	Margin %	Box
Chardonnay, Sonoma Coast	58	18	410	40.00	16,400	69%	Plowhorse
Prosecco DOC	60	12	380	48.00	18,240	80%	Star
Pinot Noir, Willamette Valley	72	24	260	48.00	12,480	67%	Star
Cabernet Sauvignon, Paso Robles	64	22	150	42.00	6,300	66%	Plowhorse
Sancerre	82	28	95	54.00	5,130	66%	Puzzle
Rioja Crianza	62	16	70	46.00	3,220	74%	Dog
Champagne brut	125	45	60	80.00	4,800	64%	Puzzle
Grüner Veltliner, Austria	68	20	45	48.00	2,160	71%	Puzzle
Barolo	150	60	40	90.00	3,600	60%	Puzzle
Malbec, Mendoza	48	13	30	35.00	1,050	73%	Dog

A few things jump out. The Chardonnay is the best seller on the list and earns less per bottle than any other popular wine. The Rioja has one of the best percentages on the page and still lands among the Dogs. And the Pinot Noir is a Star by 35 cents, so a single distributor price increase would move it into the Plowhorses without anyone noticing.

The four boxes and what to do with each wine

The box names come from food menu engineering, and they've stuck because they're easy to remember. What changes on a wine list is what you do about each one.

Margin per sale →

Puzzles Low sales, high margin. Make them visible. Stars High sales, high margin. Protect them. Dogs Low sales, low margin. Sell through, then cut. Plowhorses High sales, low margin. Reprice or recost.

Units sold →

Stars: high sales, high margin

These are the wines that pay for the list. Your job is to protect them. Never let a Star run out, because every night it's missing, the table that wanted it most likely orders something that earns you less. Keep its price and its place on the page steady, and check the distributor price every time you reorder, since a Star with a creeping cost quietly turns into a Plowhorse.

In the example, the Prosecco and the Pinot Noir are Stars. Watch the Pinot first.

Plowhorses: high sales, low margin

Guests like these wines and they earn too little. Cutting them would be a mistake, since they're often the reason a table orders a bottle at all. You have two levers instead.

The first is a small price increase. In my experience, a popular wine absorbs a modest increase far better than an unpopular one absorbs any change. Raise the Chardonnay from $58 to $62 and, at the same 410 bottles, you add $1,640 of margin a year from a single line.

The second is the cost. Negotiate with your distributor rep, ask about case or quantity discounts, or replace the wine with one of the same style that costs you less, so guests get the same experience at a better margin for you.

Plowhorses also tend to gather at the cheap end of the list for a reason. A guest who doesn't want to look cheap often orders the second-cheapest bottle, so the wine in that slot sells on its position as much as its merits. In the example, the Chardonnay sits in exactly that slot. Whatever you put there will sell, so make sure it earns.

Puzzles: low sales, high margin

Each sale of a Puzzle is worth a lot, and too few people buy it. Puzzles are usually one of two things: your most expensive bottles, or good wines your guests don't understand.

For the second kind, the fix is visibility. Give the wine a plain description of what it tastes like and which dish it suits, and a better spot on the page. Then open a bottle for your servers at pre-shift, so they can recommend it with a straight face. A Grüner Veltliner sells badly when the guest has never heard of the grape and can't pronounce it, and much better when the line tells them it's the white to order with the fish.

For the expensive bottles, look at the price. With one flat multiplier, your top bottles end up looking expensive next to the retail price a guest can find on their phone in seconds. A regressive markup, where the multiplier gets smaller as the cost goes up, earns fewer dollars per bottle and gives the bottle a real chance to sell. If the Barolo dropped from $150 to $130, each bottle would leave $70 instead of $90. If that sold 70 bottles a year instead of 40, it would earn $4,900 instead of $3,600.

Change one thing at a time and measure again after a month, so you know which change did the work.

Dogs: low sales, low margin

These wines earn little and sell little, and they hold cash on a shelf while they do it. Bottles that sit for 30 to 90 days tie up money you've already paid out, and trade sources put the spoilage risk at up to around 8% of wine inventory. The usual answer is to sell through what you have, by the glass or as a server recommendation, and not to reorder.

Before you cut, ask whether the wine has a job. It might be the only dessert wine for the dessert menu, or the one a regular asks for every Friday. A Dog with a job stays, at a price that makes sense. The rest can go, and your list gets shorter and easier for a guest to read.

In the example, the Malbec and the Rioja are Dogs, and the Rioja proves the point about percentages: 74% margin, and still not worth its place unless it has a job.

How often to run it

Once a list goes live, it tends to stay untouched. As one wine-trade writer put it, "once busy season starts, most wine programs freeze for months." Meanwhile distributor prices keep moving, and a Star can slide into the Plowhorses between two reorders.

A rhythm that works for most restaurants:

• Once a year, run the full matrix on twelve months of sales.
• Every season, run a quick check on the lines that sell most, since that's where small moves add up to real money.
• After any change, such as a new price, a new wine by the glass or a distributor price increase, re-run the lines it touches a month later.

The first time takes an afternoon, mostly spent gathering the three documents. After that the sheet is built, and an update is a matter of pasting in a new POS report.

What the matrix can't tell you

The chart tells you which wines to look at. It can't tell you what to replace a Dog with, or why a good wine stays a Puzzle. Those answers come from two places a scatter chart never sees: whether each wine sits beside a dish that actually leaves your kitchen, and whether a guest can understand the line when they read it.

The best result I can show you came from exactly those two. A 40-seat bistro I worked with roughly tripled the wine it sold after we rebuilt its selection around the food menu and replaced French names the guests couldn't read with plain descriptions that matched the dishes. Neither change would have shown up as a dot on this chart, although both would have moved the dots once the next POS report came in.

And if you'd rather do it yourself, you now have everything you need. Export the POS report this week, pull the latest invoices, and fill in the template. Then start with your best-selling wine, because if it turns out to be a Plowhorse, that one line is probably worth more than everything else on the chart.