Chapter 15: Bar and Beverage Management: Pour Cost, Inventory Control, Cocktail Programs, and Wine Lists

"The kitchen makes the reputation. The bar makes the payroll." — constructed; the thing every operator with a liquor license works out by about month four

Overview

Here is the decision on your desk. You are about to sign off on a beverage program: eight cocktails, six draft lines, a roughly forty-bottle wine list, a well, and a back bar. Somebody is going to ask you what your pour cost target is, and you will say twenty-two percent because that is what people say. Then a bartender will free-pour a half ounce over on every drink for a year and nobody will notice, because liquid does not leave a footprint on a plate and nobody counts a bar the way they count a walk-in.

So let's be precise about what is actually at stake, because it is not a rounding error.

Bellwether's plan puts beverage at 28% of revenue — \$434,000 of the \$1,550,000 year-one forecast — at a 22% pour cost. That is \$95,480 of product and **\$338,520 of contribution, or \$6,510 every week. And it is the reason the plan's blended cost of goods sold is 27.8% rather than 30%: seventy-two cents of every dollar comes in at 30% food cost, twenty-eight cents comes in at 22%, and the weighted result lands under the food-cost line. The bar is what pulls blended cost below 30%.** Take the bar away and the whole prime-cost target fails — not by a little, by 2.2 points, on a plan whose entire operating profit is sixteen points.

That is the commercial fact of this chapter. Everything else is the craft of protecting it.

The craft is unusual, because the bar is simultaneously the highest-margin station in the building and the leakiest. The product is liquid, portable, popular, and easy to give away. It is consumed by adults who like it, poured by people who are tipped by those adults, and measured — if it is measured at all — by somebody eyeballing how full a bottle looks at eleven forty at night. A kitchen that over-portions gets caught by a cost card and a scale. A bar that over-pours gets caught by nothing, unless you build the nothing into something.

We are going to cost a cocktail the way Chapter 11 costed a plate — completely, including the ice. We are going to do keg math with the foam in it. We are going to learn the tenths method and a counting cadence. And then we are going to spend the longest section of the chapter on the leaks, because that is where the money is.

Chapter 16 owns wine, which is a different business — inventory that ages, capital that sits, a service problem more than a buying problem. This chapter owns spirits, cocktails, beer, and the bar as both a station and a control problem. Chapter 8 already put Bellwether inside an open-issuance liquor jurisdiction and owns licensing and dram-shop law; we operate inside that, we do not re-teach it. And responsible service — which is not a cost topic and will never be treated as one here — gets its bar-level treatment in §15.7 and its full treatment in Chapters 18 and 25.

In this chapter, you will learn to:

• Show, in dollars, why a beverage dollar is worth more than a food dollar, and what a shift in the sales mix does to blended cost of goods sold and to prime cost.
• Compute pour cost from the usage formula, split it by category, and read the variance between what the recipes say you should have used and what the count says you did.
• Build a cocktail cost card that includes citrus at tested yield, house syrup, garnish, ice, and a spillage allowance — then price the drink and state its pour cost and contribution margin.
• Work keg math for every common keg size, quantify foam and line loss, and evaluate a line-cleaning contract on its real economics rather than its advertised ones.
• Count a bar using the tenths method, on a cadence, with a variance threshold that tells you when to look and when to leave it alone.
• Put a dollar figure on over-pour, unrecorded comps, spillage, and transfers to the kitchen — and name the specific practice that closes each one.
• Compute the break-even volume multiplier on any discount, and hold a happy-hour proposal to it.

Learning Paths

🏗️ Opening — the whole chapter, but §15.1 and the Business Plan checkpoint most of all. Your beverage assumptions carry more of the plan than you think, and §15.5's back-bar and glasswasher discussion is a build-out decision you make once. 📋 Managing — weight §15.2, §15.6, and §15.7. Pour cost by category, a counting cadence you actually keep, and the leak arithmetic are the three things that separate a bar you manage from a bar that manages you. 🍸 Beverage — all of it is yours, twice. Work §15.3 and §15.4 with a calculator and your own products. Then read Chapter 16 immediately after; wine is the half of the program this chapter deliberately leaves alone. 🚚 Small Format — beverage licensing off-premise is a different world (Chapter 8), but §15.3's costing discipline and §15.8's discount arithmetic transfer intact. A canned cocktail or a keg on a truck obeys exactly the same math with fewer places to hide.

15.1 Why beverage margin subsidizes the kitchen

Ask a chef-owner what the bar is for and you will usually get an answer about the guest experience: somewhere to wait, somewhere for the solo diner, a cocktail that matches the food. All true. Now ask the accountant, and you get a different and equally true answer: the bar is a margin subsidy that lets the kitchen cook the way the chef wants to cook.

Both answers describe the same twelve seats.

The arithmetic, in one paragraph

A dollar of food sales at Bellwether's 30% food-cost target leaves 70 cents of gross contribution. A dollar of beverage sales at the 22% pour-cost target leaves 78 cents. The difference is eight cents on every beverage dollar, and Bellwether plans \$434,000 of beverage dollars. Eight cents on \$434,000 is **\$34,720 a year** — which is not a nice extra. It is the difference between a plan that hits its prime-cost target and one that does not.

FIGURE 15.1 — Two dollars, two destinations                  [the Bellwether plan — constructed]

  A FOOD DOLLAR                          A BEVERAGE DOLLAR
  ─────────────────────────────          ─────────────────────────────
  product cost   ██████         30¢      product cost   ████           22¢
  contribution   ██████████████ 70¢      contribution   ████████████████ 78¢

  THE BLEND:   72% food × 30.0%  =  21.60¢
               28% bev  × 22.0%  =   6.16¢
               ───────────────────────────
               BLENDED COGS      =  27.76¢  →  27.8%

    food only .......... 30.0%   ────────────────────  the food-cost line
    with the bar ....... 27.8%   ◄── the bar pulls you 2.2 points under it

  2.2 points on $1,550,000 of sales = $34,720 a year.

  PRIME COST with the bar:     27.8% COGS + 32.3% labor = 60.1%   on target
  PRIME COST without it:       30.0% COGS + 32.3% labor = 62.3%   over

Read the bottom two lines again, because they are the chapter in miniature. Chapter 1 established prime cost at or below 60% as the full-service benchmark, and Bellwether's plan lands at 60.1% — on the line, with no room. Remove the beverage subsidy and the same restaurant, cooking the same food, paying the same people, runs 62.3%. Nothing about the kitchen changed. The bar changed.

This is why an operator who says "I'm not really a bar person, we'll just keep it simple" has made a financial decision without noticing. Simple is fine. Uncosted is not.

🧮 Run the Numbers

What \$12.88 a head is actually doing.

Bellwether's dinner check is \$46.00**, split **\$33.12 food / \$12.88 beverage. That 28/72 split is not a description of guest behavior. It is a target, and every decision in this chapter exists to serve it.

On plan, dinner runs about 95 covers a night, five nights, fifty-two weeks — 24,700 dinner covers a year. At \$12.88 each, that is **\$318,136 of dinner beverage. The remaining \$115,864** of the \$434,000 comes from brunch, the patio in season, and private events.

Now watch what a four-point miss on attachment does. Suppose beverage lands at 24% instead of 28%, with food per guest unchanged at \$33.12. Then beverage per guest is \$10.46, and the check falls to \$43.58** — down **\$2.42.

\$2.42 × 24,700 dinner covers = **\$59,774 of revenue gone, of which 78 cents on the dollar — \$46,624** — was contribution. That is **\$897 a week**, on a restaurant whose entire beverage program contributes \$6,510 a week.

A four-point miss on a mix assumption nobody stress-tested costs you fourteen percent of the bar's annual contribution. This is why "we'll figure out the drinks later" is an expensive sentence.

Why the margin is structurally better, and where the structure ends

Beverage costs less as a percentage of price for four honest reasons, and one dishonest one.

One: the product arrives finished. A case of wine does not need to be butchered, and a bottle of rye does not need a yield test the way a whole fish does. There is no trim loss, no cooking loss, no overnight spoilage on a sealed bottle. Chapter 11's whole apparatus of as-purchased versus edible-portion cost mostly collapses to a single number: cost per ounce.

Two: the portion is exact and small. A drink is one and a half to two ounces of the expensive thing. The rest is ice, water, citrus, and a mixer that costs pennies. A plate is six ounces of the expensive thing and there is no way to make it four.

Three: shelf life is long. Spirits do not spoil. Unopened beer and most wine hold for months. Compare that to a case of berries that has a four-day window. Waste, which is a permanent tax on food cost, is close to optional on the spirits line.

Four: the guest's price expectation is anchored to the experience, not the ingredient. People who would balk at a \$29 chicken will pay \$15 for four ounces of liquid, because they are buying the room, the craft, and the occasion. This is not a con; it is the same thing the dining room sells. But it is why beverage pricing has more headroom than food pricing.

And the dishonest reason: nobody counts it. A bar that has never been counted always reports a better pour cost than it has, because the product that walked out the door was never in the numerator. Half of the "great" pour costs in this industry are measurement artifacts. We fix that in §15.2 and §15.6.

The structure ends in three predictable places, and you should know them before you build a program around the margin:

• Wine is the exception. A wine list typically runs 25–35% cost — meaningfully worse than spirits, and worse than beer in many programs — because bottle pricing is transparent to guests and because inventory sits. Chapter 16 works this properly. For now, note only that a beverage program weighted toward wine is a lower-margin program than one weighted toward cocktails, and that this is a choice, not an accident.
• Beer, especially craft draft, is not the cheap line people assume. §15.4 shows a keg running 28% after foam.
• Premium spirits invert the percentage. You cannot mark a \$70 bottle up six times. A two-ounce pour of something rare might run a 30% cost and still be one of the best contribution-margin dollars on the back bar. Chapter 12's lesson holds exactly: you bank dollars, not percentages.

🤝 Hospitality

The twelve seats that do more than twelve seats' worth of work.

Bellwether's bar is twelve of sixty-eight seats — 17.6% of the room. It is also the part of the room that behaves least like the rest of it.

The bar seats the walk-in the reservation book cannot place. It seats the solo diner who would otherwise occupy a two-top for ninety minutes and spend forty dollars. It seats the couple who arrived twenty minutes early and would otherwise be standing in the vestibule deciding they should have gone somewhere else. It turns faster than a table because bar guests self-select for shorter stays. And it sells more beverage per cover than a table does, because the product is eighteen inches away and the person making it is looking at you.

Chapter 24 measures this properly, in revenue per available seat-hour. Chapter 22 owns the seating decision. What matters here is that the beverage line and the seating strategy are the same decision, and a restaurant that treats its bar as a waiting room has thrown away its highest-yield seats.

One practical consequence: whoever is behind that bar is not a beverage-dispensing appliance. On a Tuesday they are the entire hospitality experience for four solo guests, and on a Friday they are the first person forty people speak to. Staff and pay the position accordingly. Chapter 17 makes the turnover argument; it applies to bartenders more than to almost anyone, because a bartender who knows forty regulars by name is carrying revenue you cannot replace by posting a job.

15.2 Pour cost: computing it, targeting it, and reading the variance

Pour cost is beverage cost of goods sold divided by beverage sales, expressed as a percentage. It is the bar's version of food cost percentage, and it lies in all the same ways plus two new ones.

$$\text{Pour cost \%} = \frac{\text{Beverage COGS}}{\text{Beverage sales}}$$

And beverage COGS is not what you bought. It is what you used, which Chapter 13 gave you as the usage formula and which does not change at the bar:

$$\text{Usage} = \text{Beginning inventory} + \text{Purchases} - \text{Ending inventory}$$

If you have not counted your ending inventory, you do not have a pour cost. You have a ratio of invoices to sales, which moves every time a distributor delivery lands on the wrong side of a period close. Buy three cases of wine on the thirtieth and your "pour cost" spikes. Sell down the back bar and it magically improves. Neither event has anything to do with how you poured.

Split it, always

Here is the first thing that separates a bar manager from someone who quotes a number. An aggregate pour cost is a blend of four different businesses, and reading only the blend will send you to the wrong place every time.

Bellwether's plan, split:

(All figures the Bellwether plan; constructed and internally consistent. Category targets are consistent with common industry ranges — spirits in the mid-teens to low twenties, draft beer in the low-to-mid twenties, package beer higher, wine in the high twenties to mid thirties — but ranges vary enormously by market, list, and pricing philosophy. Build your own from your own cost cards.)

Now consider a month where the aggregate pour cost prints 23.4% instead of 22.0%. Alarming. Except that a cold February pushed the mix toward wine and away from patio cocktails, and wine costs 28% while cocktails cost 17%. The mix moved. The discipline did not. You can produce a worse pour cost from a better-run month, and if you do not split the number you will spend a week interrogating a bartender who did nothing wrong.

Split it and the picture is unambiguous: if spirits held at 17.0% and wine held at 28.0%, the categories are fine and the blend moved. If spirits printed 19.5%, you have something to find.

Define the sub-metric while we are here. Liquor cost is spirits COGS divided by spirits sales — the spirits line alone, with no wine, beer, or soda in it. When a bar manager says "my liquor cost is sixteen," that is the number they mean, and it is not the same number as pour cost. Know which one you are quoting and to whom.

Ideal versus actual, applied to liquid

Chapter 11 built the ideal-versus-actual comparison for food: what your recipes say you should have used, against what your count says you did. The same discipline applies to liquid, and at a bar it is easier to run and more informative, because the recipes are shorter and the units are exact.

To compute ideal usage you need three things: a cost card for every drink you sell, a point-of-sale system that reports units sold by item, and somebody who re-costs when a price moves. Multiply units by recipe cost, sum, and divide by sales. That is your ideal pour cost. The difference between it and actual is the variance, and the variance is the chapter.

Most modern point-of-sale and inventory systems will compute a "theoretical" beverage cost for you. Use it — and understand its one fatal dependency: it is only as good as the recipes loaded into it. A bar that changed its house syrup in March and never updated the cards is comparing this month's reality to last year's fiction. Re-costing is a calendar item, not an inspiration.

Reading the variance

Variance thresholds for a bar Bellwether's size — roughly \$8,400 of beverage sales a week:

That last row deserves emphasis and it is the most useful thing in this section. A large variance is much more often a counting error, a missed invoice, a keg nobody logged, or a transfer to the kitchen that was never booked than it is theft. Recount. Check the invoice file. Ask the sous chef what went into the braise. Only then do you have a control question, and Chapter 34 owns what happens next.

⚠️ Where the Money Leaks

Two silent distortions that live inside every pour cost.

The transfer. Cooking wine, brandy for the sauce, beer for the mussels, bitters in a vinaigrette, the bottle of rye that goes into the dessert. All of it is beverage inventory that becomes food cost, and if nobody books the transfer it stays in beverage COGS against beverage sales that never happened. At Bellwether, an illustrative \$40 a week of product moving from bar to kitchen is \$2,080 a year — 0.48 points of pour cost. It also understates food cost by the same amount, so you get to be wrong twice.

The fix costs nothing: a transfer sheet on a clipboard by the walk-in door. Date, item, quantity, who took it. The bar manager totals it weekly and the number moves from one column to the other. Ten minutes a week to stop being half a point wrong all year.

The denominator. Does your "beverage sales" include coffee, tea, soda, and the zero-proof list? Bellwether's does — \$26,040 of it, at an 11.5% cost. Including cheap non-alcoholic sales in the denominator makes the blended pour cost look better than the liquor discipline actually is. Compute and report all five: liquor cost, wine cost, beer cost, non-alcoholic cost, and the blend. Then when someone asks you what your pour cost is, ask them which one they want.

🔍 Check Your Understanding

1. A bar begins the week with \$11,200 of inventory, purchases \$1,640, and ends with \$10,980. Beverage sales were \$8,100. What is the actual pour cost?
2. Your aggregate pour cost improved from 22.4% to 21.1% with no change in procedure. Give two explanations that have nothing to do with better bartending.
3. Why is a 2.5-point variance a reason to recount rather than a reason to investigate a person?

(1: usage = 11,200 + 1,640 − 10,980 = \$1,860; 1,860 ÷ 8,100 = 23.0%. 2: The sales mix shifted toward low-cost categories — more cocktails and soda, less wine; or a large delivery landed after the count, deflating usage this period and inflating it next. Either produces the movement with identical discipline. 3: Because large variances are far more often measurement failures — a missed invoice, an uncounted keg, an unbooked transfer, a mis-entered price — than losses, and an accusation you have to retract costs more than the variance did.)

15.3 Costing a cocktail, line by line, including ice, garnish, and dilution

Chapter 11 built the recipe cost card for food and established the discipline: every component, costed at its real yield, plus a waste allowance, summed to a plate cost, divided by a target to get a price. We apply exactly that method to liquid. Nothing about the technique changes. What changes is the list of things people forget.

People forget citrus at yield. They forget house syrups have an ingredient cost and a labor cost. They forget the garnish. And essentially everyone forgets the ice — which is fine on any one drink and is \$509 a year at Bellwether's volume.

The drink

Bellwether's signature cocktail is the Rivermill Sour: two ounces of house rye, a quarter ounce of amaro, three quarters of an ounce of fresh lemon, three quarters of an ounce of house demerara syrup, two dashes of aromatic bitters, shaken hard and served up in a chilled coupe with an expressed lemon peel and a sprig of thyme.

Here is what it costs.

🧾 Read the Numbers

```text FIGURE 15.2 — "The Rivermill Sour" [the Bellwether plan — constructed; all purchase prices illustrative] THE ARTIFACT Cocktail cost card, house signature drink, priced at $15.00 on the bar menu. THE CONTEXT Bellwether, opening season. Purchase prices are the plan's assumed landed cost from a single distributor. They are ILLUSTRATIVE — spirits pricing varies enormously by state, by distribution structure, and by volume, and in control states it is set by the state. Cost your own products.

COMPONENT PURCHASE UNIT COST/UNIT AMOUNT LINE ───────────────────────────────────────────────────────────────────────────────── Rye whiskey, house 750 mL = 25.4 fl oz @ $27.94 $1.100/oz 2.00 oz $2.20 Amaro (modifier) 750 mL = 25.4 fl oz @ $30.48 $1.200/oz 0.25 oz $0.30 Lemon juice, fresh lemon @ $0.35, yield 1.1 oz $0.318/oz 0.75 oz $0.24 Demerara syrup 2:1 2 lb @ $1.92/lb → 32 oz batch $0.120/oz 0.75 oz $0.09 Aromatic bitters 5 oz bottle @ $8.00 $1.600/oz 2 dashes $0.10 Garnish lemon ÷ 4 peels = $0.09; thyme bunch $2.40 ÷ 120 = $0.02 — 1 each $0.11 Ice $1,650/yr operating cost ÷ ~41,000 ice services — ~8 oz $0.04 Service items napkin, pick — 1 $0.03 ───────────────────────────────────────────────────────────────────────────────── COMPONENT SUBTOTAL $3.11 Spillage, breakage, and remakes @ 3% $0.09 ───────────────────────────────────────────────────────────────────────────────── TOTAL DRINK COST $3.20

MENU PRICE $15.00 POUR COST $3.20 ÷ $15.00 21.3% CONTRIBUTION MARGIN $15.00 − $3.20 $11.80

FINISHED VOLUME 3.78 oz of ingredients + ~22% shake dilution (0.83 oz) = 4.61 oz into a 7 oz coupe

WHAT IT SHOWS A completely costed drink at 21.3% — above the 17.0% spirits-category target, and that is correct. A craft cocktail with four ingredients and a garnish will always cost more than a well highball. The category average is held down by the $12 rye on the rocks at 14.0% and the $10 vodka soda at 5.8%. What you bank is the $11.80. WHAT IT DOESN'T It does not include the bartender's twenty seconds, the eleven minutes of prep labor in every batch of syrup, the juicing shift, the electricity in the glass chiller, or the glassware that breaks. A cost card prices ingredients, not drinks. It also assumes the recipe is followed — which is an assumption, not a fact, and §15.7 is about what it costs when it isn't. THE DECISION Price at $15.00 and hold it. Load the recipe into the POS at $3.20 so the theoretical usage report is real. Re-cost when the rye contract renews or when lemons cross $0.45, whichever comes first. THE LESSON Cost the whole drink or don't bother. The eight cents of ice, garnish, and napkin look ignorable on one card and are $1,017 a year across the cocktail list — and the four items people usually omit are precisely the four that never appear on an invoice line you would notice. ```

The lines people get wrong

Bottle conversions. Get these right once and stop guessing.

(One US fluid ounce is 29.5735 mL. These are theoretical maximums — zero spillage, zero drip, zero heel left in the bottle. Nobody achieves them.)

Citrus at tested yield. This is Chapter 11's yield test, run on fruit. A lemon at \$0.35 that yields 1.1 ounces of juice costs \$0.318 an ounce — but a cold lemon straight from the walk-in yields closer to 0.8 ounces, which is \$0.44 an ounce, a 38% increase you will never see on an invoice. Room-temperature fruit, rolled before juicing, is a cost control. So is juicing to a par rather than juicing a case because it is Tuesday: fresh citrus juice holds about a day for lemon and less for lime before it tastes like a different drink.

House syrups have two costs. The demerara syrup above costs \$0.12 an ounce in sugar. It also costs about eleven minutes of a prep hand's time per 32-ounce batch, which at a loaded \$19 an hour is \$3.48 — more than the sugar. At Bellwether's volume that labor sits in the kitchen's prep schedule (Chapter 14) rather than on the cost card, which is a defensible convention as long as you say so out loud. What is not defensible is pretending it is free. If a syrup takes forty minutes, either batch it bigger or buy it.

Garnish is inventory. Cut citrus is a par item with a shelf life measured in hours. Herb sprigs die. A garnish tray that gets built for a 200-cover Saturday and thrown out on a 60-cover Tuesday is waste with a name, and it belongs on the same waste log Chapter 13 built for food.

Ice and dilution. These are two separate ideas that live in the same sentence, and both matter.

Ice is a cost. An ice machine consumes electricity, water, sewer, filter cartridges, and a service visit twice a year. Bellwether's illustrative all-in operating cost is \$1,650 a year, spread across roughly 41,000 ice services — water glasses, cocktails, and iced non-alcoholic drinks — which is four cents each. Across 12,718 cocktails a year that is \$509. It is the smallest line on the card and the one most cost cards omit entirely, and it is exactly the kind of omission that joins four others to become three points.

Dilution is a specification. A hard shake adds roughly 20–25% of the drink's volume in water; a stir adds 15–20%. The Rivermill Sour's 3.78 ounces of ingredients finish at about 4.61 ounces. That water is free volume the guest experiences as drink — which is a gift to your margin — but it is also the reason a cocktail recipe is a specification and not a suggestion. A bartender who shakes eight seconds instead of fifteen serves a smaller, stronger, differently-balanced drink. On Tuesday the guest thinks it was great. On Friday the guest thinks it was different. Consistency is the product.

👨‍🍳 On the Line

Eight-forty on a Friday, and where the cost card breaks.

The cost card assumes a clean sequence: jigger, shake, strain, express, garnish, sell. Here is what is actually happening at 8:40.

There are four tickets in the service well, three guests standing behind the twelve seats waiting for a stool, and a server who needs two glasses of the by-the-glass red that the bartender has to reach past the ice bin to get. The dishwasher has not sent glassware up in eleven minutes. The bartender is making a round of four Rivermill Sours and one is a substitution — no thyme, allergy.

Watch the cost card break in four places. The bartender free-pours the fourth rye because the jigger is under a shaker tin. One shake gets over-diluted because it sat while a card was run. A coupe goes over on the drain board and takes a second glass with it. And the allergy drink gets re-made because the first one got a thyme sprig out of muscle memory.

None of that is a discipline problem. All of it is a design problem, and the fixes are physical: a jigger in a dedicated well that nothing else occupies, two shaker sets so one is always clean, glassware par depth that survives a fifteen-minute dish gap, a printed modifier that reads ALLERGY-NO THYME in a way that stops a hand, and a barback on Friday and Saturday whose entire job is that the bartender never leaves the station.

The 3% spillage allowance on the card exists for exactly this Friday. If your allowance is 3% and your actual is 6%, the answer is almost never "tell the bartender to be careful." It is that the station is fighting them. Chapter 7 designed the bar; this is where you find out whether the design was right.

Pricing from the card

Chapter 11's target-cost pricing works unchanged: divide the drink cost by the target percentage.

$$\text{Price} = \frac{\text{Drink cost}}{\text{Target pour cost}} = \frac{\$3.20}{0.20} = \$16.00$$

At a 20% target the Rivermill Sour prices at \$16. Bellwether prices it at **\$15**, which is a 21.3% cost, and does so deliberately for three reasons that will recur throughout this book:

1. Menu psychology and the price ladder. Chapter 10 established that a menu is read as a set, not as items. A cocktail list that runs \$14–\$16 reads differently in the Rivermill District than one that runs \$15–\$18, and the signature drink is the anchor.
2. You bank dollars. \$11.80 of contribution on a drink that sells sixty times a week beats \$12.80 on a drink that sells thirty. Chapter 12's argument, applied to liquid.
3. The percentage is a servant. Chapter 11 said to price from a target and then know when to ignore the answer. A one-dollar deviation on the house cocktail, taken knowingly, is fine. A one-dollar deviation nobody computed is how cost drift starts.

What you may not do is price a cocktail list by looking at what the place down the street charges. That is not pricing; that is copying somebody else's cost structure and hoping it resembles yours.

15.4 Draft beer: keg math, yield, foam loss, and line cleaning as a cost control

Draft beer is the most operationally demanding product on the bar and the one operators most consistently assume is easy. It arrives in a container you cannot see into, it is served through sixty feet of tubing at a temperature and pressure you have to actively maintain, and a meaningful fraction of it leaves as foam.

Keg math is the arithmetic that converts a keg into sellable glasses. Start with the container.

(All theoretical, at zero loss. One US gallon is 128 fluid ounces.)

Bellwether runs six draft lines — a house lager on half barrels and five rotating craft taps on sixth barrels — with a standard pour of 14 ounces of beer in a 16-ounce glass, finished with roughly an inch of head. That last detail is a pricing decision as much as a service one: a "pint" in an American restaurant is almost never sixteen ounces of liquid, and an operator who prices as though it is has given away two ounces a glass.

Draft yield: what you actually get

Draft yield is sellable ounces divided by keg ounces. The losses are unglamorous and they are constant: the first pour of each shift after the lines warmed overnight, the foam on a fresh keg before it settles, the mid-shift foaming when a cooler cycles, the pour that gets dumped because it came out ninety percent head, and the last two inches of a blowing keg that nobody can serve.

Industry rule of thumb is 8–15% loss on a well-maintained system. Bellwether's plan uses a measured 12%.

FIGURE 15.3 — Where a sixth-barrel keg actually goes       [constructed teaching example;
                                                             keg prices are ILLUSTRATIVE]

  Sixth barrel (sixtel), 5.16 gal × 128 oz         =  660.5 oz   ████████████████████
  Pour standard: 14 oz of beer in a 16 oz glass
  ───────────────────────────────────────────────────────────────────────────────────
  Theoretical glasses         660.5 ÷ 14           =   47.2      ████████████████████
  Less first-pour foam, cooler-cycle foam, dumped
  pours, and the tail of a blown keg   (12%)       =   -5.7      ██
  ───────────────────────────────────────────────────────────────────────────────────
  SELLABLE GLASSES                                 =   41.5      █████████████████

  Keg delivered cost (illustrative)                    $105.00
  Cost per THEORETICAL glass    $105.00 ÷ 47.2     =     $2.22   →  24.7% at $9.00
  Cost per SELLABLE glass       $105.00 ÷ 41.5     =     $2.53   →  28.1% at $9.00
                                                                    ─────────────
  Foam is worth 3.4 points of pour cost on this keg — $51.30 of beer
  you paid for, carried in, tapped, chilled, and poured down a drain.

Now compare the same math on a half barrel of the house lager at an illustrative \$145 delivered: 1,984 ounces, 141.7 theoretical glasses, 124.7 sellable at the same 12% loss, \$1.16 a glass. At \$7.00 that is a 16.6% pour cost.

Sit with those two numbers. \$2.53 a glass and \$1.16 a glass, same bar, same night, same bartender. Nothing about staff discipline explains the difference; the keg size, the beer, and the price do. This is the most important thing to understand about draft: your beer pour cost is mostly a purchasing and menu decision, and only secondarily an execution one. A tap list of six sixth-barrel craft beers priced like domestic lager is a structurally unprofitable beer program that no amount of careful pouring will rescue.

Bellwether's plan lands the draft line at 21.7% because it blends the two — roughly 25 half barrels and 73 sixth barrels a year, about 98 kegs, two a week — and prices the craft taps at \$9 while the house lager sits at \$7. Package beer (bottles and cans) runs 31.7% on \$17,360 of sales, which is normal and which is why the whole beer category lands at 24.2% rather than in the teens.

🧮 Run the Numbers

Does line cleaning pay for itself? Only if you count honestly.

Draft lines grow yeast, bacteria, mold, and beer stone. The industry standard is a full chemical cleaning every two weeks. Bellwether's six lines, cleaned by an outside service at an illustrative \$10 a line, cost **\$60 a cleaning × 26 cleanings = \$1,560 a year.**

The yield argument, run properly. Bellwether needs 6,127 sellable glasses a year (\$52,080 of draft revenue at a blended \$8.50). At 12% loss that requires 97,475 ounces of beer, costing \$11,290 — the 21.7% draft cost. Let the lines go dirty and loss rises to 20%. To sell the same 6,127 glasses you now need 107,223 ounces — 9,748 extra ounces at \$0.116 an ounce = \$1,129 a year of extra beer.

\$1,129 of benefit against \$1,560 of cost. On yield alone, line cleaning does not pay.

That is the honest answer and almost nobody gives it. Here is why you clean them anyway.

• Remakes. Dirty lines pour foam and off-flavors. Two remade or sent-back beers a service across 364 services is 728 glasses of product at \$2.53 — **\$1,842 a year. Add that and the cleaning pays roughly 1.9 to 1**.
• The guest who stops ordering draft. A guest served a sour, buttery, or cloudy beer does not complain. They order a cocktail next time, or a bottle, or they go somewhere else. This is unquantifiable and it is larger than the other two combined.
• Sanitation. Beer lines are food-contact surfaces. Chapter 25 owns the inspection; an inspector who asks for your line-cleaning log and is handed nothing has learned something about your building.
• The in-house alternative. Chemical, a pump, and ninety minutes of a bar manager's time runs about \$36 a cleaning — **\$936 a year** — at which point even the yield-only argument is close to break-even and the full argument is overwhelming. The catch is that in-house cleaning done badly is worse than none, and it has to survive turnover.

The lesson generalizes past beer. A control that does not pay for itself on the obvious number may still be the cheapest thing you do, and an operator who only funds measures with clean paybacks will underinvest in exactly the ones that protect revenue rather than reduce cost.

The rest of the draft system, briefly

Three more things decide your draft yield, and all three are Chapter 7 decisions you now live with:

Temperature. Beer served above roughly 38°F foams. The walk-in that holds the kegs, the trunk line, and the faucet all have to stay cold, and a long-draw system needs glycol. A cooler that cycles to 44°F on a hot Saturday will foam every pour during your busiest hour.

Pressure and balance. The applied CO₂ pressure has to balance the resistance of the line — its length, diameter, and rise. Too little and the beer pours flat and foamy; too much and it over-carbonates in the keg over days and pours foamy in a different way. A "balanced" system is a plumbing calculation, and it is worth paying a draft technician to do it once.

Beer-clean glassware. A beer-clean glass is one with no residual fat, protein, or detergent film — head forms and laces down the side as the glass empties. A glass washed with kitchen detergent and dried with a towel is not beer-clean, and beer poured into it collapses. This matters at Bellwether specifically, and §15.5 explains why.

15.5 Building a bar program: well, call, premium, and the shape of the back bar

A back bar is a menu that a guest reads before anyone hands them anything. It is also a working station where somebody has to make four drinks in ninety seconds. Those two jobs pull in different directions and the resolution is a tiering decision.

The three tiers of the pour

• Well (also "rail," "house," or "speed rail") — the spirit poured when a guest does not name a brand: "vodka soda," "whiskey ginger." It lives in the speed rail at the bartender's hip, it is the highest-volume and lowest-cost product on the bar, and it is invisible to most guests.
• Call — the guest names the brand. Mid-tier, priced a dollar or two above well, and where most of the spirits volume actually sits in a restaurant like Bellwether.
• Premium (also "top shelf") — higher-priced brands, often sipped neat or on the rocks, usually displayed on the back bar's upper tiers where guests can read them.

A standard pour is the specified volume for a given drink type, written into the recipe and trained. Bellwether's standards:

Write them down. Post them. A standard nobody can produce in writing is a preference, and preferences drift.

The premium-well decision, priced

Every operator faces this one and most decide it by feel. Do it with arithmetic instead.

An illustrative cheap well vodka at \$18.00 per 1.75 liters is \$0.304 an ounce. A "premium well" — a genuinely decent bottle used as the default — at \$26.00 per 1.75 liters is \$0.439 an ounce. On a 1.5-ounce pour the difference is \$0.20 a drink.

Bellwether pours roughly 3,815 well drinks a year (about 30% of its 12,718 spirit-based drinks). The upgrade costs \$763 a year — about \$15 a week.

Now the other half, which most people skip. If the better well supports a house highball at \$11 instead of \$10:

The pour cost gets worse and you make \$0.80 more per drink — \$3,052 a year on 3,815 drinks. This is Chapter 12's lesson in its purest form: the percentage went the wrong way and the dollars went the right way, and the dollars are what you deposit. An operator managing to a pour-cost percentage alone would reject this trade, which is precisely why nobody should manage to a pour-cost percentage alone.

Note also what the table conceals: a \$10 vodka soda at a 5.8% pour cost is, in percentage terms, the single most profitable item in the building, and nobody has ever put it on a menu. Some of the best margin in a restaurant is in products the guest orders without looking.

The shape of the back bar

FIGURE 15.4 — Bellwether's bar: the working geometry    [the Bellwether plan — schematic,
                                                          not to scale]

  BACK BAR — guest-facing display and storage
  ┌───────────────────────────────────────────────────────────────────────────┐
  │ TIER 3   premium / sipping spirits, amari — the bottles guests read        │
  │ TIER 2   call brands — the ones ordered by name                           │
  │ TIER 1   backups, house wine, syrups, glass storage                       │
  │          [ GLASS CHILLER ]              [ under-bar cooler: beer / wine ]  │
  └───────────────────────────────────────────────────────────────────────────┘
        ↑ 36 in. of bartender aisle — nothing stored on the floor, ever

  ┌───────────────────────────────────────────────────────────────────────────┐
  │ ┌───────────┐   ┌────────────────────┐   ┌──────────┐   ┌───────────────┐ │
  │ │ SERVICE   │   │  MAIN WELL         │   │  DRAFT   │   │  DIRTY GLASS  │ │
  │ │  WELL     │   │  speed rail (8)    │   │  6 taps  │   │     DROP      │ │
  │ │ rail (6)  │   │  ice bin · 3-comp  │   │  drip    │   │  → runs to    │ │
  │ │ ice · gun │   │  sink · garnish    │   │  tray    │   │    dish pit   │ │
  │ └───────────┘   └────────────────────┘   └──────────┘   └───────────────┘ │
  │  ← servers        ← the bartender lives here →            ← barback       │
  └───────────────────────────────────────────────────────────────────────────┘
     BAR TOP — 12 seats   ●  ●  ●  ●  ●  ●  ●  ●  ●  ●  ●  ●
                          └──────────── guests ────────────┘

  LEGEND:  ● = bar stool   │ = counter edge   ← = traffic direction
  NOTE:    There is NO under-counter glasswasher in year one — Chapter 7
           deferred it. Dirty glass travels. See below.

Four principles are doing the work in that diagram, and they are worth stating because they are transferable to any bar you will ever build or inherit.

One: the service well is separate from the guest well. Servers ringing in drinks for the dining room must not stand in the bartender's guest-facing workspace. A bar without a dedicated service well forces every server order through the same eighteen inches of counter the guests are sitting at, and the result on a Friday is a bartender who is rude to somebody through no fault of their own.

Two: everything the bartender touches most is within one step. Speed rail at the hip, ice bin under the working hand, garnish tray above it, jigger and tins in a dedicated slot. Chapter 7 called this workflow and counted steps; the same arithmetic applies here. Two extra steps per drink, twelve thousand drinks a year, is a labor number.

Three: the back bar is tiered for both jobs. Premium up where guests read it, call in the middle where the bartender reaches it, backups and working goods low. Tier 3 is merchandising. Tier 2 is operations.

Four: dirty glass leaves the station immediately. Which brings us to the thing Bellwether does not have.

👨‍🍳 On the Line

Living without the glasswasher for a year.

Chapter 7's build-out deferred the under-counter glasswasher to year two. That was a real decision with a real number attached: an installed cost of roughly \$7,400**, against **\$1,273 a year of extra labor — about fifteen minutes a service of barback time running glass to the dish pit and back, which is 91 hours a year at a loaded \$14 an hour. Payback: 5.8 years. On that arithmetic alone, deferring is correct.

The arithmetic alone is not the whole story, and you should know what you bought with the \$7,400 you did not spend.

• Deeper glass par. Glass in transit is glass you do not have. Bellwether has to carry roughly eight extra dozen glasses in circulation — an illustrative \$264 of smallwares and shelf space it did not plan for.
• More breakage. Every extra handling event is a breakage opportunity. Glass gets carried, racked, run, racked again, and carried back.
• A ceiling at 9:30 on Saturday. The constraint on a busy bar is rarely the bartender's hands. It is ice, and it is glassware. A fifteen-minute gap in the dish return is a fifteen-minute gap in the bar's throughput, and you will not see it on any report.
• Beer-clean glassware becomes a policy, not a machine setting. Kitchen dish detergent leaves a film that kills head retention and makes good draft beer look and taste flat. The three-compartment bar sink with a proper glass brush and a bar-specific detergent is not optional now — it is the only beer-clean pathway the building has, and it has to be trained and audited.

What a good operator does with a constraint like this: name it, price it, work around it deliberately, and put the fix on a dated list rather than a wish list. Buy the machine in year two. Until then, the glass protocol is a written standard, the barback is scheduled on Friday and Saturday specifically to protect it, and the bar manager checks glassware for film once a week. A constraint you have decided to accept is a plan. A constraint you keep re-discovering at 9:30 on Saturday is a problem.

Program size and the discipline of a short list

One more decision, and it is the same one Chapter 10 made about the menu. Every additional bottle on the back bar is inventory, capital, a par level, a count line, and a chance to be out of stock on something you promised.

Bellwether's program: eight cocktails, six draft lines, a well and a modest call selection, a working premium tier, and roughly forty wines (Chapter 16). Eight cocktails is a deliberate number. It fits on a single menu panel, it can be batched partially in prep, a new bartender can learn it in two shifts, and every ingredient on it is used by at least two drinks — Chapter 10's cross-utilization argument applied to the bar, where it matters more than in the kitchen because bar ingredients have shorter open lives and no second use as staff meal.

A back bar with 180 bottles on it and forty that move is not a program. It is a museum with a liquidity problem, and it turns up in §15.6 as a large ending-inventory number that never gets smaller.

15.6 Bar inventory: the tenths method, weighing, and counting cadence

You cannot compute a pour cost without an ending inventory, and a bar ends every period with sixty to a hundred and twenty partially full bottles. So the entire practical problem of bar inventory reduces to one question: how do you count a bottle that is not full?

The tenths method

The tenths method is the industry's answer and it is exactly what it sounds like. You look at each open bottle, judge how full it is to the nearest tenth, and write down a decimal. A bottle about three-quarters full is 0.8. One with a couple of ounces left is 0.1. Then:

$$\text{Bottles on hand} = \text{full bottles} + \sum \text{tenths}$$

Multiply by the bottle cost to get a dollar value.

Worked, for one product: four sealed bottles of house rye at \$27.94, plus one on the rail reading 0.4 and one on the back bar reading 0.7.

$$4 + 0.4 + 0.7 = 5.1 \text{ bottles} \times \$27.94 = \$142.49$$

That is the whole technique. Its virtues are that it is fast — a trained pair counts Bellwether's entire bar in about forty-five minutes — that it requires no equipment, and that its errors are random rather than systematic, so across sixty bottles they largely cancel.

Its limits, stated plainly:

• A single bottle is accurate to maybe ±5%, and worse on dark glass, embossed bottles, and odd shapes. On one bottle of amaro that is fifteen cents. Across a whole count it averages out. Do not chase a two-tenths disagreement.
• It systematically flatters the neck and the base. Bottles are not cylinders. A bottle that looks half full usually has less than half in it, because the shoulder holds volume the eye overestimates. Some operators correct by counting the shoulder generously and the base conservatively; most just accept it as a constant bias that cancels between periods, which it does as long as the same person counts the same way.
• It cannot detect dilution. If someone waters a bottle, tenths will never find it. That is a Chapter 34 problem and it needs a Chapter 34 tool.
• Consistency beats accuracy. A count that is 3% high every single week produces a correct usage figure, because the bias appears in both the beginning and ending inventory and cancels in the subtraction. A count that is 3% high one week and 1% low the next produces a variance that is entirely fictional and that you will spend a Monday investigating.

The alternatives, and when they earn their keep

Weighing. Put the bottle on a scale, subtract the known empty-bottle weight, and convert weight to volume by the product's density. Accurate to roughly ±1–2%, and it is the method behind most inventory apps that pair with a Bluetooth scale. It is slower per bottle and it needs a tare weight for every bottle shape you stock. Worth it for high-value products and for any bar where the variance is already under investigation.

Gauge sticks and marked bottles. A ruler calibrated for a specific bottle shape, or a bottle marked at each ounce. Cheap, faster than weighing, more accurate than eyeballing, and useless the moment a distributor changes the bottle mold.

Metered pour spouts and automated systems. These record every pour electronically. They are accurate, they eliminate the counting problem for spirits entirely, and they cost real money and change the bar's character — both for the staff, who read them as distrust, and for guests, who can see them. They belong in high-volume, high-turnover, low-supervision operations. At a twelve-seat chef-driven bar they would be a strange choice and would probably cost you the bartender you want.

The point-of-sale as a cross-check. Your POS knows exactly how many Rivermill Sours you sold. Two ounces each. If the POS says 61 and your count says 134 ounces of rye left the shelf, you have a question, and you did not need a scale to find it.

The cadence

A counting schedule that nobody keeps is worth less than an honest weekly estimate. Design for adherence.

Five rules make counts usable:

1. Same time, always. After close, before deliveries. Bellwether counts Sunday night because the week runs Tuesday through Sunday and the bar is closed Monday.
2. Same order, always. Walk the bar the same direction, the same shelf sequence, every week. The count sheet is printed in that order. This is the single biggest driver of speed and consistency.
3. Two people: one counts, one writes. Faster, more accurate, and it means no single person is the sole author of the number.
4. Count the whole beverage inventory, not just the fun parts. Spirits, wine, beer in every location, mixers, syrups, juice, garnish. The storeroom is where the surprises live.
5. Value it consistently. Use the same cost per unit until it changes on an invoice, and change it everywhere on the same day.

One capital note that surprises operators the first time they see it. Bellwether carries roughly \$11,800** of beverage inventory against **\$95,480 of annual beverage COGS, which is about eight turns a year — 45 days of supply. The kitchen turns its walk-in every week to ten days. Beverage is slow-turning capital, wine most of all, and it will be one of the larger single line items in the opening inventory. That is not a criticism of the product; it is the reason a bar program should be short, deliberate, and free of bottles you bought because a rep was persuasive.

🧾 Read the Numbers

```text FIGURE 15.5 — "Week 19: the bar count" [the Bellwether plan — constructed] THE ARTIFACT Weekly beverage inventory and usage worksheet, Tuesday open through Sunday close, week 19. Counted at 11:40 p.m. Sunday by the bar manager and one bartender, tenths method, same sheet and same order as every other week. Ideal usage pulled from the POS item-mix report against loaded cost cards. THE CONTEXT Late spring. Patio open two weeks. No menu changes, no price changes, no new staff behind the bar. One bartender was on vacation and a part-timer covered three shifts.

                            BEGIN    + PURCH    - END      = USAGE    IDEAL   VAR
Spirits                    $4,180      $590     $4,122       $648     $572   +$76
Wine                        5,910       840      5,848        902      895     +7
Beer (draft + package)      1,260       410      1,329        341      326    +15
Non-alcoholic and mixers      470        75        484         61       58     +3
────────────────────────────────────────────────────────────────────────────────
TOTAL                     $11,820    $1,915    $11,783     $1,952   $1,851  +$101

Beverage sales, week 19                                     $8,410
ACTUAL pour cost      $1,952 ÷ $8,410                        23.2%
IDEAL  pour cost      $1,851 ÷ $8,410                        22.0%
VARIANCE                                                   +1.2 pts

WHAT IT SHOWS One category is the whole story. Wine (+$7), beer (+$15), and non-alcoholic (+$3) are inside counting noise. Spirits ran $648 against a $572 ideal — 13.3% over, and 75% of the total variance. At a blended $1.20 an ounce across the spirits inventory, $76 is about 63 extra ounces on roughly 245 spirit-based drinks: a quarter ounce a drink. That is the arithmetic signature of free pouring, not of theft. WHAT IT DOESN'T It does not prove anything. $101 could be a mis-tenthed amaro, a keg logged to the wrong week, a comp nobody rang, an invoice not yet entered, or three shifts of a part-timer who has never been trained on the jigger standard. It also says nothing about WHICH spirits — the sheet is valued by category here; the line-item detail is on the back page and that is where you actually look. THE DECISION Three things, in order, this week. (1) Pull the line detail and find which three products carry the $76. (2) Re-train and re-check the jigger standard with the part-timer before assuming anything about anyone. (3) Recount the spirits shelf Wednesday morning against Sunday's sheet to confirm the count itself. Do not raise this at a staff meeting yet — you have one week of data. THE LESSON $101 a week is $5,252 a year, and it arrives one quarter ounce at a time in a bar where every single person is doing their honest best. This is why you count weekly instead of monthly: at four weeks you would be looking at $404 and four candidate explanations, and by then the habit is set. ```

15.7 The leaks: over-pour, free pour, comps, spillage, and sweethearting

This is the heart of the chapter.

Food leaks in ways you can see. An over-portioned plate looks over-portioned. A case of lettuce that went bad smells. A cook who is stealing has to carry something out of the building.

Liquid leaks invisibly. A half ounce is a splash. It looks like generosity, it tastes like hospitality, it produces a better tip, it costs the guest nothing, and the bottle it came out of will be counted by somebody eyeballing tenths at eleven forty at night. There is no plate to look at, no smell, nothing to carry out. This is why bars are the most-controlled station in most restaurants and still the leakiest — the product is designed to be given away.

Start with the arithmetic, because until it is a number nobody funds the prevention.

Over-pour: the half ounce

Over-pour is pouring more than the standard, whether from habit, generosity, speed, or inattention. It is almost never deliberate theft and it is almost always in the same direction: over. Nobody free-pours short on purpose. Every incentive at the bar — the tip, the guest's face, the bartender's own pride in a generous drink — points one way.

⚠️ Where the Money Leaks

What a half ounce costs, across a year.

Bellwether sells about 12,718 spirit-based drinks a year — \$173,600 of spirits and cocktail revenue at a blended \$13.65 average, roughly 245 drinks a week. The blended cost of the spirits inventory is about \$1.20 an ounce.

Add a half ounce to every one of them:

• 12,718 drinks × 0.5 oz = 6,359 ounces of spirit
• 6,359 oz × \$1.20 = **\$7,631 a year**
• 6,359 oz ÷ 25.4 oz per 750 mL bottle = 250 bottles — nearly five a week
• Pour cost moves from 22.0% to 23.8% — 1.76 points

\$7,631 is more than a full week of the bar's entire contribution (\$6,510), given away half an ounce at a time by people who think they are being nice.

A quarter ounce — which is what Bellwether's week-19 count actually showed — is \$3,815 a year and 0.88 points. Half the leak, still four times the annual cost of every jigger, spout, scale, and count sheet in the building.

Now run it the other way, on a single bottle. House rye at \$27.94, poured neat at a \$12 price:

	Standard 2.0 oz pour	Free-poured 2.25 oz
Pours per 750 mL (25.4 oz)	12.7	11.3
Revenue per bottle	\$152.40 | \$135.60
Pour cost on that bottle	18.3%	20.6%

A quarter ounce of drift gives away \$16.80 a bottle — more than half of what the bottle cost you. Do it on the 1.5-ounce highball instead, where a quarter ounce is a sixth of the pour, and the damage is proportionally worse.

Jigger versus free pour

Free pour is pouring by count — the "four-count," where the bartender counts while the bottle is inverted — with no measure. Jiggering is pouring into a measured vessel and then into the tin or glass.

The case for free pour is real and you should hear it fairly. It is faster. It looks better. A practiced bartender with a consistent spout on a consistent bottle can hold a count within a respectable tolerance. And on a twenty-deep Saturday it is genuinely quicker.

The case against it is arithmetic. A count depends on the spout, the fill level of the bottle, the bottle's shape, how tired the person is, how loud the room is, and whether they are talking to a guest while pouring — which they always are, because that is the job. Every one of those variables moves the count up.

🧮 Run the Numbers

The jigger costs \$233 a year and saves \$7,631.

Jiggering adds roughly three seconds per drink. Across Bellwether's 12,718 spirit-based drinks that is 38,154 seconds — 10.6 hours a year. At a loaded \$22 an hour for a bartender, the jigger policy costs \$233 a year in labor.

It closes a leak worth \$7,631** at a half ounce, or **\$3,815 at a quarter.

Even at the conservative figure that is a sixteen-to-one return on the most boring policy in the building.

Now the limits, because a technique sold without them is a sales pitch.

• Three seconds is not the Saturday constraint. Glassware, ice, and the service well are. If you remove the jigger to go faster and the bottleneck was the dish pit, you gave away \$7,631 for nothing.
• A jigger does not stop deliberate theft. Someone who intends to give drinks away will jigger perfectly and not ring the sale. That is Chapter 34's problem and it needs Chapter 34's tools.
• A jigger used badly is theater. Jiggering and then topping off "a little more" is a slower way to over-pour. The standard is the jigger and nothing after it.
• Some guests read a measure as stinginess. The answer is a correctly sized, correctly priced, correctly made drink and a bartender who neither explains nor apologizes. A properly built cocktail does not feel small. A weak one does, and weak usually comes from bad ratios, not from an honest two ounces.

The workable policy at a restaurant bar: jigger everything that goes into a cocktail, jigger every premium pour, and permit a trained count-pour only on the well highball — then audit that permission with the POS cross-check in §15.6. If the audit drifts, the permission ends.

Comps, buybacks, and the difference between hospitality and a hole

A comp is a drink given away deliberately. Chapter 31 owns comps, voids, and discounts as accounting objects; Chapter 34 owns the authorization controls. What belongs here is the bar's specific version, because the bar is where comping is a culture, not an exception.

Buying a round for a regular, sending a drink to a guest whose entrée was late, a shift drink for the kitchen after a brutal Saturday, a taste of an amaro to sell a bottle of it — these are real hospitality tools with real returns. Chapter 23 makes the case for service recovery and it is a good one. None of that is the problem. The problem is that a comp with no budget, no authorization, and no record is indistinguishable from a loss, and after a while it becomes one.

Price it. Three comped drinks a service, seven services a week:

• 3 × 7 × 52 = 1,092 drinks a year
• At the \$13.65 blended price: **\$14,906 of revenue given away**
• At \$3.20 of product: **\$3,494 of cost**
• Lost contribution: \$11,433 a year — 1.76 weeks of the bar's entire annual contribution

Now the subtle part, and it is the most useful thing in this section. Recording the comp does not save you a dollar. If the drink is rung and comped, gross sales rise by \$14,906 and a comp line takes it back out; net sales, COGS, and pour cost are identical to the unrecorded case. The money is exactly the same.

What recording buys you is that you can see it. An unrecorded comp is a pour-cost variance with no explanation, indistinguishable from over-pour, spillage, or theft, and you will spend three Mondays looking for it. A recorded comp is a line on a report with a name, a reason code, and a total, and the conversation becomes "we comped \$287 last week, here is who and why, is that the number we want?" — which is a management conversation instead of an investigation.

Comp and spill tracking is the practice: every pour that does not become a sale gets rung into the POS with a reason code — comp, buyback, spill, remake, staff tasting, transfer. The product leaves inventory and leaves a record. It takes four seconds. Set a written policy covering who may comp, how much per shift, and what has to be entered; Chapter 34 builds the authorization architecture.

Spillage is the honest cousin: the dropped glass, the over-shaken drink, the pour that came out foam, the cocktail sent back, the bottle that hits the floor. Bellwether's cost cards carry a 3% allowance. If actual spillage runs 5%, that is \$1,835 a year — and, per the On the Line callout in §15.3, the answer is usually a station design problem rather than a carelessness problem. Log it and you will find out which.

Sweethearting is the last one and it needs to be named accurately. It is the practice of giving free or under-rung drinks to friends, regulars, or industry guests, typically in exchange for a larger tip. It is the most common form of bar loss in the industry, and it very rarely begins as theft. It begins as hospitality that nobody attached a budget to. Chapter 34 owns the definitions, the detection patterns, and the response. What belongs here is the operating truth: the line between generous and gone is authorization and recording, and it is your job to draw it in writing before anyone crosses it, not after. A bartender who has been told "take care of people" and given no number has been set up to fail by a manager who wanted to sound generous.

FIGURE 15.6 — The bar's leak ladder, annualized          [the Bellwether plan — constructed]

  LEAK                                        ANNUAL      POUR-COST
                                               COST         POINTS
  ───────────────────────────────────────────────────────────────────────────
  Free-pour drift, ½ oz per drink             $7,631        1.76   ████████
    (at ¼ oz per drink)                       $3,815        0.88   ████
  Comps at 3 per service, unrecorded          $3,494        0.81   ████
    (revenue given away: $14,906)
  Untracked transfers to the kitchen          $2,080        0.48   ██
  Spillage at 5% against a 3% allowance       $1,835        0.42   ██
  Draft loss at 20% instead of 12%            $1,129        0.26   █
  ───────────────────────────────────────────────────────────────────────────
  ALL FIVE AT ONCE (using the ½ oz figure)   $16,169        3.72

  Pour cost:      22.0%  →  25.7%
  Blended COGS:   27.8%  →  28.8%    (0.72 × 30.0% + 0.28 × 25.7%)
  PRIME COST:     60.1%  →  61.1%    over target, on the bar alone

  $16,169 is 4.8% of the bar's entire annual contribution of $338,520,
  and not one dollar of it involves anybody stealing anything.

Read the bottom block twice. The bar, by itself, running five ordinary undisciplined habits, can push the entire restaurant past its prime-cost target. Not the kitchen. Not labor. The bar — which was supposed to be the thing subsidizing everyone else.

That is the argument for every count sheet, jigger, transfer log, and comp code in this chapter, and it is why the bar gets a control system rather than a suggestion.

⚖️ Code and Compliance

Responsible service is not a cost-control topic, and it never becomes one.

Everything above this callout is about protecting margin. This is not, and the two must never be confused. A beverage program that pushes volume without a service standard that limits it is a liability generator with a good pour cost.

Chapter 8 established Bellwether's licensing and dram-shop exposure; Chapter 18 owns the training program and Chapter 25 owns certification. What belongs at the bar, on the shift:

• Over-service is the bartender's hardest judgment and the building's largest single exposure. In many states, dram shop statutes expose the business — and in some states the individual server — to civil liability for injuries caused by a guest who was served while visibly intoxicated, and for service to a minor. The rules, the standards of proof, and the caps vary enormously by state. Verify locally, with counsel.
• Track pace, not just count. Three drinks in forty minutes is a different situation from three over two hours. A bartender who is watching only the tab is not watching the guest.
• Check identification on a written standard — everyone who appears under a stated age, every time, including people you know — and train the staff on what a valid ID actually looks like in your state and on what to do with one that isn't.
• Write the refusal protocol down before you need it. Who makes the call, what gets said, what the alternative is (water, food, a ride, a call to whoever is with them), and how it goes into the shift log. Document it the same night, every time.
• Back the refusal, in public. A manager who reverses a bartender's cutoff to keep a guest happy has just taught the whole staff never to make the call again. This is the single most consequential cultural decision a bar manager makes, and it takes about four seconds.
• Read your incentives. If your compensation, your promotions, and your pre-shift meeting all reward volume, and your written policy limits it, the incentives win. Design the program so they point the same way: never run a promotion whose mechanic rewards speed of consumption, and never make a bartender choose between the standard and the sales target.

Nothing in the rest of this chapter — not the jigger, not the pour cost, not the happy hour — outranks this.

15.8 Happy hour, promotions, and the arithmetic of a discount

Somebody is going to propose a happy hour. It will sound obviously good: the bar is empty at five, rent is being paid anyway, a full room begets a full room. Some of that is true. All of it is testable, and almost nobody tests it.

Here is the test.

The break-even multiplier

A discount does not reduce your price by the discount. It reduces your contribution margin by the discount, and contribution margin is a much smaller number than price. That is the whole insight, and it is why discounts are so much more expensive than they feel.

$$\text{Break-even volume multiplier} = \frac{\text{Original CM}}{\text{Discounted CM}} = \frac{P - C}{P_d - C}$$

Run it on the Rivermill Sour: price \$15.00, cost \$3.20, contribution margin \$11.80.

A one-third discount requires you to sell seventy-three percent more drinks to earn exactly what you were earning before. Not to gain — to stand still. Halve the price and you need nearly two and a half times the volume.

The same math on the \$9.00 craft draft at \$2.53 sellable cost: contribution \$6.47. At a \$6.00 happy-hour price the contribution is \$3.47 and the multiplier is 1.865 — you need 86.5% more beer. Beer is worse than cocktails because its cost is a larger fraction of its price, which is the general rule: the higher the product cost percentage, the more brutal the discount. Never discount your worst-margin category first, which is exactly what most operators do because it is the cheapest-sounding thing to give away.

🧮 Run the Numbers

Bellwether's happy-hour proposal, priced honestly.

The proposal. Tuesday through Friday, 4:30–6:00 p.m., bar and patio only. \$10 cocktails (from \$15), \$6 drafts (from \$9), \$8 wines by the glass (from \$12), and a \$9 bar snack.

What the window does today. The bar is close to empty. Illustratively, \$210 a service across four services — \$840 a week**, roughly 60% beverage and 40% food. COGS: \$504 × 22% = \$110.88 plus \$336 × 30% = \$100.80, so \$211.68**. Contribution: **\$628.32 a week.** A bartender is already on the clock setting up, so there is no incremental labor.

What the proposal projects, per service:

	Units	Price	Sales	Unit cost	COGS
Cocktails	26	\$10.00 | \$260.00	\$3.20 | \$83.20
Draft beer	18	\$6.00 | \$108.00	\$2.53 | \$45.54
Wine by the glass	10	\$8.00 | \$80.00	\$2.40 | \$24.00
Bar snack	16	\$9.00 | \$144.00	\$2.70 | \$43.20
Per service			\$592.00** | | **\$195.94

Contribution per service: \$592.00 − \$195.94 = \$396.06.** Four services: **\$1,584.24 a week.

Incremental labor. A second bartender and a prep hand for the snacks, about \$52 a service — \$208 a week.

Net contribution with happy hour: \$1,584.24 − \$208.00 = \$1,376.24 a week. Net contribution today: \$628.32. Gain: \$747.92 a week — call it \$38,900 a year.

That is a real number and it is a good one, and it comes almost entirely from the fact that the rent, the insurance, the manager's salary, and the building are already paid for during that window. Chapter 24 calls these shoulder hours and Chapter 32 explains why filling them is disproportionately profitable: the fixed costs are sunk, so nearly all of the contribution drops through.

Now the assumption that decides the whole thing: are these guests incremental?

Suppose 6 of the 26 cocktails a service would have been sold at \$15.00 later that evening. Each one now earns \$6.80 instead of \$11.80 — a loss of \$5.00. Six a service, four services, fifty-two weeks: **\$6,240 a year.** The gain falls to \$32,660.

Suppose 12 of the 26 would have come anyway: **\$12,480 a year**, and the gain falls to \$26,420.

The program is still worth doing in all three cases — but notice that cannibalization, not pricing, is the variable that moves the answer, and it is the one nobody measures. Measure it: track the 4:30–6:00 window and the 7:00–9:00 window, together, for eight weeks against the same weeks last year or the same weeks in a comparable period. If the late window falls, you moved revenue rather than created it.

The five failure modes of a promotion

One: it trains the guest to wait. A guest who learns that the \$15 cocktail is \$10 at 5:30 will never again buy it at 6:30 if they can help it. Time-limited discounts are permanent price signals. This is the argument for keeping the happy-hour window genuinely early and genuinely short.

Two: it shifts the mix toward your worst margins. If the discount is deepest on the item with the highest cost percentage — usually beer — you have engineered a mix shift into your least profitable product, and your blended pour cost will get worse even if every individual decision was defensible.

Three: it changes who is in the room. This is a positioning question, not a math question, and Chapter 2 owns it. A deep-discount happy hour brings a different guest than a \$46 dinner check does. Sometimes that is exactly the point — trial, neighborhood presence, filling a dead window. Sometimes it quietly repositions a restaurant its owners still describe as chef-driven. Decide which on purpose.

Four: it strains an operation that was staffed for empty. Twenty-two covers in the bar at 5:15 with one bartender who is also setting up for dinner is a bad experience for everyone, and a happy hour that produces bad service in the window before your dinner service is actively expensive.

Five: the responsible-service failure mode, which is the serious one. A promotion whose mechanic rewards drinking quickly — anything expiring at a hard time, anything priced per-round, anything that makes a second drink cheaper than the first — pushes consumption into a compressed window. That is a real over-service risk and it is a real liability. Structure discounts as a fixed price on a defined list, available for a defined window, with food available throughout, and hold exactly the same service standards you hold at nine o'clock. A guest at a discounted price is not a different guest.

⚖️ Code and Compliance

Happy hour is regulated, and the rules are wildly inconsistent.

Alcohol regulation in the United States is a state matter and the variation is larger than in almost any other area of restaurant compliance. On discounting alone, states differ on all of the following:

• Some states prohibit happy hour outright or ban reduced drink prices for limited periods.
• Some permit discounts but prohibit "two-for-one," unlimited, or all-you-can-drink formats.
• Some require that discounted drinks be available for a full day rather than a window, on the theory that a deadline encourages rapid consumption.
• Some require food service alongside discounted alcohol.
• Some restrict advertising of drink prices, on-premise or off.
• Hours of service, last call, and Sunday rules vary by state and frequently by county or city.

There is no way to write a chapter that tells you what your rule is. Verify locally, in writing, with your state and local alcohol authority and with counsel, before you print a menu. A promotion that violates a liquor regulation puts at risk the single most valuable and least replaceable asset the restaurant owns — and Chapter 8 already established what a liquor license is worth and how long it takes to get back.

A practical habit: keep a one-page compliance summary for your jurisdiction — service hours, ID standard, training requirement, discount rules, advertising rules — posted where the bar manager can see it, and re-check it annually. Rules change and nobody sends you a letter.

🔍 Check Your Understanding

1. A cocktail sells for \$14 and costs \$3.50. You discount it to \$9. How many more units must you sell to break even on contribution?
2. Why does the same percentage discount hurt a draft beer more than a cocktail?
3. Your happy hour produced \$700 a week of additional contribution. What single measurement would most change your confidence that the number is real?

(1: CM goes from \$10.50 to \$5.50; multiplier = 10.50 ÷ 5.50 = 1.909, so +90.9% units. 2: Because the beer's cost is a larger share of its price, so the same dollar discount removes a larger share of a smaller contribution margin. 3: The sales in the hours immediately AFTER the promotion window, compared to a comparable prior period — if the 7:00–9:00 window fell, you moved revenue rather than created it.)

🍽️ The Business Plan

Checkpoint 15 of 40 — the Beverage Program.

Chapter 11 costed the food menu. This checkpoint contributes the beverage half of the same document: what Bellwether pours, what it costs, what it earns, and the control system that keeps the earning real. Chapter 16 adds the wine section immediately after this one.

The beverage line, stated as a target rather than a hope.

The category plan and its targets:

The program. Eight cocktails on a single panel, anchored by the Rivermill Sour at \$15.00 — \$3.20 cost, 21.3% pour cost, \$11.80 contribution. Six draft lines: one house lager on half barrels at \$7.00 and five rotating craft taps on sixth barrels at \$9.00, at a 14-ounce standard pour in a 16-ounce glass. A premium well, priced for at \$11.00 on the house highball. A short call and premium tier. Roughly forty wines, which Chapter 16 builds.

The control system, which is the part that makes the target real:

1. Jigger standard, written and posted, on every cocktail and every premium pour; a permitted trained count-pour on the well highball only, audited monthly against the POS item mix.
2. Weekly count Sunday after close, tenths method, two people, same order, ~45 minutes. Monthly full count including storeroom, with a second manager present.
3. Ideal-versus-actual by category every week. Investigate at 1.0 point or any category more than 10% over its ideal, two weeks running. Recount before concluding anything.
4. Comp and spill tracking: every non-sale pour rung with a reason code. A written comp allowance per shift. (Chapter 34 builds the authorization architecture.)
5. Transfer sheet by the walk-in for everything moving from bar to kitchen — an estimated \$40 a week that otherwise lands as 0.48 points of phantom pour cost.
6. Keg log at every tap and blow; draft yield tracked per keg; line cleaning every two weeks at an illustrative \$1,560 a year, funded on quality and sanitation grounds rather than on yield alone.
7. Responsible service: written refusal protocol, a posted ID standard, pace tracking, shift-log documentation, and a stated rule that management backs a bartender's cutoff. Training and certification per Chapters 18 and 25.

The year-one constraint we are carrying. No under-counter glasswasher until year two — Chapter 7's deferral, at \$1,273 a year of extra barback labor (about 15 minutes a service, 91 hours a year) against a 5.8-year payback on a roughly \$7,400 install. The plan accepts it, funds an extra eight dozen glasses (about \$264), schedules a barback Friday and Saturday specifically to protect glass flow, and writes a beer-clean glassware protocol for the three-compartment bar sink. Revisit in month fourteen.

The happy-hour proposal, included with its hurdle attached. Tuesday–Friday 4:30–6:00, bar and patio, at a projected \$747.92 a week of additional contribution — about \$38,900 a year — subject to (a) local regulatory verification before a menu is printed, (b) an eight-week cannibalization measurement of the 7:00–9:00 window, and (c) a structure that never rewards speed of consumption.

What this checkpoint does not settle — and it is a longer list than the settled one.

• Wine. Thirty-eight percent of the beverage line and the worst pour cost in the program is entirely Chapter 16's, including list architecture, by-the-glass yield, preservation, and the capital sitting in bottles.
• Whether 28% attachment is real. It is an assumption about guest behavior, not a fact. §15.1 showed that a four-point miss is \$46,624 of contribution — \$897 a week. This is the largest single unverified number in the beverage plan.
• Whether the market pays \$15 for a cocktail. Chapter 2 sized the trade area; nobody has yet bought a drink in it.
• What the pour cost actually is once real bartenders pour real drinks on real Fridays. The plan says 22.0%. Week 19 of the constructed count already says 23.2%.
• The controls. Everything above is a system; Chapter 34 owns separation of duties, cash handling, the daily sales report, and what you do when the variance is a person rather than a procedure.
• Seasonality. Patio beverage is concentrated in perhaps twenty weeks and the February beverage line will look nothing like the June one. Chapter 33 owns what that does to cash.

Open questions carried forward:

1. Does the Rivermill District actually buy \$15 cocktails at 28% of check? (Chapters 22, 24, 27)
2. Does the wine list at 28% cost pull the blended pour cost above target, and does the by-the-glass program hold its yield? (Chapter 16)
3. Will a weekly count survive contact with a busy season, or will it become a monthly count in October? (Chapters 21, 34)
4. What is the real draft yield once six lines run in a hot kitchen's ambient heat all summer? (Chapters 25, 34)
5. Does the happy hour create covers or move them? (Chapters 24, 27, 32)

Conclusion

The bar is the best margin in the building and the easiest place to lose it, and both facts come from the same property: the product is liquid.

Beverage runs a 22% cost against food's 30%, which pulls Bellwether's blended cost of goods sold to 27.8% and its prime cost to 60.1% — on the benchmark, with nothing to spare. Remove the bar's subsidy and the same restaurant runs 62.3%. The bar is what holds the plan on its line, which is why \$12.88 of beverage per dinner guest is a target and not an observation, and why every decision in this chapter — the cost card, the jigger, the count sheet, the transfer log — exists to protect it.

The method is the one Chapter 11 already taught you, applied to liquid without apology. Cost the whole drink, including the citrus at tested yield, the house syrup, the garnish, and the four cents of ice. Price from a target and then decide knowingly when to deviate. Count what you actually used, by category, every week. Compare it to what the recipes say you should have used, and treat the difference as information rather than as an accusation.

And then take the leaks seriously, because they are the largest quantifiable opportunity in this chapter. A half ounce of habitual over-pour is \$7,631 a year — more than a full week of the bar's entire contribution. Five ordinary undisciplined habits together are \$16,169, enough to push the whole restaurant past its prime-cost target without a single person stealing anything. The jigger that closes most of it costs \$233 a year in labor. That is the arithmetic, and once you have seen it you cannot un-see it.

Two things this chapter deliberately did not do. It did not teach wine, because a wine list is a different business — aging inventory, sitting capital, and a selling problem more than a buying one — and Chapter 16 takes it up next. And it did not treat responsible service as a cost topic, because it is not one. Over-service, identification, pace, and the refusal that gets backed by management sit above every number in this chapter, and a program whose incentives quietly push against them has been designed badly no matter what its pour cost prints.

Key Terms

Pour cost — beverage cost of goods sold divided by beverage sales, expressed as a percentage. The bar's counterpart to food cost percentage; computed from counted usage, never from invoices alone. Bellwether targets 22.0%. (Ch. 15)

Liquor cost — spirits cost of goods sold divided by spirits sales alone, excluding wine, beer, and non-alcoholic. A sub-metric of pour cost and frequently confused with it. (Ch. 15)

Standard pour — the specified volume of a spirit for a given drink type, written into the recipe and trained: at Bellwether, 1.5 oz for a highball, 2.0 oz for a neat pour, and per-recipe amounts in cocktails. (Ch. 15)

Free pour — pouring by count, without a measure. Faster and theatrical; it drifts upward because every incentive at the bar points that direction. (Ch. 15)

Jigger — a measured pour vessel, and by extension the policy of measuring every pour. Adds roughly three seconds a drink and closes most of the over-pour leak. (Ch. 15)

Well (also rail, house) — the default spirit poured when a guest does not name a brand; the highest-volume, lowest-cost product on the bar. (Ch. 15)

Call — a spirit the guest orders by brand name; the middle tier of the back bar and where most restaurant spirits volume sits. (Ch. 15)

Premium (also top shelf) — higher-priced brands, often sipped neat, usually displayed on the upper back bar. Runs a higher cost percentage and often an excellent contribution margin. (Ch. 15)

Cocktail costing — building a complete cost card for a drink: spirit, modifiers, citrus at tested yield, house syrups, bitters, garnish, ice, service items, and a spillage allowance, summed to a drink cost and divided by price to give pour cost. (Ch. 15)

Draft yield — sellable ounces divided by keg ounces, after foam, dumped pours, and the tail of a blown keg. A well-run system loses 8–15%; Bellwether plans 12%. (Ch. 15)

Keg math — the arithmetic converting keg volume into sellable glasses at a stated pour size: a sixth barrel is 5.16 gal = 660.5 oz = 47.2 theoretical 14-oz glasses, or 41.5 after 12% loss. (Ch. 15)

Beer-clean glassware — glass free of fat, protein, and detergent film, on which head forms and laces down the side. A dish-machine-washed glass is usually not beer-clean, which collapses the head on otherwise good draft. (Ch. 15)

The tenths method — counting partially full bottles by eye to the nearest tenth and summing the decimals with the full bottles. Fast, equipment-free, accurate to roughly ±5% per bottle, and reliable in aggregate provided the same person counts the same way every period. (Ch. 15)

Over-pour — pouring more than the standard, from habit, generosity, speed, or inattention. Almost always upward; a half ounce a drink is \$7,631 a year at Bellwether's volume. (Ch. 15)

Comp and spill tracking — ringing every pour that does not become a sale into the point-of-sale with a reason code, so the product leaves inventory and leaves a record. It does not save money; it makes the money visible. (Ch. 15)

Beverage transfer — product moving from the bar to the kitchen (cooking wine, brandy for a sauce, beer for mussels) or the reverse, booked out of beverage cost and into food cost. Untracked, it inflates pour cost and understates food cost simultaneously. (Ch. 15)

Happy-hour economics — the arithmetic of a beverage discount: a price cut removes a share of contribution margin, not of price, so the break-even volume multiplier is (original CM ÷ discounted CM). A one-third discount on a \$15 cocktail requires 73.5% more units to stand still. (Ch. 15)

Spaced Review

1. Without looking back: state the formula for pour cost, and explain why a bar that has never counted its ending inventory does not have one.
2. Bellwether's blended cost of goods sold is 27.8% on a 72/28 food/beverage mix. Show the arithmetic, then state what blended COGS and prime cost would be if beverage fell to 20% of sales with all other targets unchanged.
3. A sixth-barrel keg costs \$105 delivered and you pour 14 ounces in a 16-ounce glass at \$9.00. What is your pour cost at zero loss, and at a 12% loss? Which number should appear on the beverage plan, and why?
4. Reaching back to Chapter 11: the Hearth Chicken costs \$8.52 and sells for \$29.00 — a 29.4% food cost and \$20.48 of contribution. The Rivermill Sour costs \$3.20 and sells for \$15.00 — a 21.3% pour cost and \$11.80 of contribution. Which item does more for the business, and what additional information do you need before answering?
5. The recurring question: a bar manager proposes removing jiggers on Friday and Saturday to speed up service. Using the numbers in §15.7, state the annual cost of the proposal, the one operating condition under which it might nonetheless be correct, and the measurement you would run before deciding.