Bar Profit Playbook: How to plug the leaks costing profits

The Money Leaving Before You See It

You know your sales number, and you can compare it to last year and last month. What no report shows you is what you are losing on the cost side: liquor poured over the standard measure, food priced without a recipe cost behind it, vendor invoices nobody checks line by line, and product that leaves the building without being paid for.

A bar with no cost controls in place typically loses 8 to 14 percent of its sales to problems that could have been prevented. That loss is not caused by a bad location or by slow nights. It happens because nobody is measuring. You can get that money back, and the rest of this page explains how.

The Four Places It Disappears

It is rarely one single problem. A bar that is losing money while its sales look healthy usually has all four of the leaks below running at the same time. Any one of them costs you real money. All four together is why the money is gone and nobody in the building can explain where it went.

What One Point Is Worth, On Your Numbers

Pour cost is usually where the biggest recoverable number lives, and the fastest to move once the controls are in. Measured pours and weekly counting consistently pull back 3 to 5 points in 60 to 90 days. Put your own sales in and see what that is worth.

What 90 Days Looks Like

This is a worked example, not a promise. Take a full-service bar and kitchen doing about 1.1 million a year, 720,000 bar and 380,000 food. The owner has run it six years. He knows the room. He does not know his pour cost, his recipe costs, or his prime cost within four points. Here is the arc the first 90 days tends to follow.

The first real measurement is almost always worse than the owner’s estimate, and it is uncomfortable to see. It does not mean anything has gone wrong. It means these are the first accurate numbers the bar has had in six years.

Week two: a signed pour policy goes up and measured pours start. Weekly counting begins. The first delivery audit catches 840 in price variances and the credits get requested. Recipe costs go in on the top ten menu items. The first variance report flags well vodka and house tequila running 11 to 13 percent over.

Week six: pour cost is at 27.8 percent, down 3.6 points. The well variance resolves to a mix of free-pouring and one Friday bartender pocketing 60 to 80 a shift in no-ring cash. Documented. Addressed. Recipe costs are in on 22 items. Four price below their cost floor: two get a price move, two get a recipe change.

Day 90: the first full quarter. Pour cost 24.3, down 7.1 points. Food cost 32.8, down 5.4. Vendor credits recovered: 6,200 in the quarter. Prime cost 59.4, under 60 for the first time in the bar's history. It is the same room, the same staff and no new customers. The only thing that changed is that these systems now run every week.

Recovery Has a Timeline

The first 30 days are about getting your starting numbers, not results. Those first numbers are your baseline, which is what everything later gets measured against. You run a real count and read your actual pour cost. It usually comes out higher than you expected. That is normal, and it is the point of the exercise: you are seeing the real number instead of the one you assumed.

Days 30 to 60 are where the first real movement happens. Measured pours are in. Counts are running on a schedule. The variance report is flagging items to chase. Pour cost usually starts moving here, a point or two at first, more if free-pouring was bad. Food cost moves slower because recipe costs take time to build, so expect food results in the 60 to 90 day window.

By day 90 every system should be running. Pour cost trending to target, food cost measurably below the day-1 baseline, the first vendor credits requested, prime cost reviewed every week. The operation looks different from the inside.

Most bars stop somewhere between day 30 and day 45, when the first burst of discipline fades and one busy week pushes the count off schedule. Whatever you use to run this, you need something that tells you a system has stopped running before the month ends.

How Controlled Are You Right Now?

Ten questions. Yes or no, no partial credit. If the answer is kind of, or we used to, or I need to check, that is a No. The system is either running this week or it is not.

Most operators know last Saturday's sales to the dollar. Ask for last week's pour cost and the room goes quiet. Ask for last month's prime cost and the subject changes. That gap is where this diagnostic lives. The monthly figure next to each No is an example of what that gap commonly costs. It is not a measurement of your bar.

Every Leak: Where It Hides, What Closes It

Before the systems themselves, here is the whole picture on one screen. For each leak: the work that captures it, the number that exposes it, and the system on this page that closes it.

Every form and policy these systems need is on this page as well. The pour standards policy, the theft and loss policy, the corrective action form, the portion audit sheet, the food handling standards and the vendor terms checklist each sit inside the system that uses them, and you can read, copy or print any of them for free.

The Numbers to Run Against

These are your reference points. Find your category, know your target, and know the line where a number turns into a problem. Put your own figure in below and the table will tell you where you sit.

Pour Cost by Category

Food Cost by Category

Prime Cost by Concept Type

Void, Comp, and Variance Red Flags

Six Systems, One Profit Machine

These are not independent fixes. Fix your pour cost and watch food cost drift up. Stop a theft problem and miss the vendor overcharges on every delivery. Get prime cost right for one quarter and lose it the next because nothing was written down and the manager who built it left. Cost control fails when you treat these as five separate problems, because each one feeds the next.

The six systems are sequenced on purpose. Pour cost comes first because it is the foundation: you cannot work out a real prime cost without a real cost of goods sold, and you cannot work that out without a weekly count. Each system feeds the next.

Pour Cost Control

A bar in Nashville, 180 seats, sports concept, about 900,000 a year in beverage. Open four years, never a real count. The owner figured pour cost was around 23 percent. Everything got counted and run: actual pour cost 34 percent. On 900,000 in bar sales that gap is 99,000 a year. Not stolen, not fraud. Just never counted.

Your POS is a sales tool. It records what was rung up. It has no way of knowing that a bartender poured 2.1 ounces into a drink specced at 1.5, or that a comp went out with nobody’s name on it, or that a bottle left the building in somebody’s bag. Pour cost is how you measure those things, and you need both. Most bars with no system in place run 6 to 12 percentage points higher than they should.

A 0.3 ounce average overage sounds like nothing. On a well spirit it is about 27 cents a drink. At 250 drinks a night, 300 nights a year, that is roughly 20,000 a year from over-pouring alone, and that is the conservative version. It is not malice. A bartender in the weeds at 10pm is thinking about the eight tickets on the rail, not the extra quarter ounce. The problem is a system that lets accuracy depend on attention during the busiest hour of the night.

Run It Without Software

A clipboard and your invoices are enough to get a true number. The arithmetic is not hard. The discipline is.

1. Count everything, the same way every week. Same day, same time, same order, same two people, before or after service but never during. Count full bottles as whole units and estimate open bottles by eye, to the nearest tenth. Being consistent matters more than being exact. If the same person estimates the same way every week, the change from one week to the next is reliable, and that change is what you are looking for. Trying to be perfectly precise is the main reason people give up on counting every week.
2. Price the count. Extend every line at your current cost per unit. That total is your opening inventory value. Keep the sheet.
3. Add the period's purchases. Every beverage invoice between the two counts, at what you actually paid, including any that arrived and were not yet entered.
4. Count again at the end of the period and price it. That is your closing inventory value.
5. Usage = opening + purchases minus closing. That dollar figure is what left the shelves.
6. Pour cost = usage divided by beverage sales for the exact same period. Net sales, after comps and discounts, never register totals. Multiply by 100.
7. Do it by category, not as one number. Spirits, draft, bottled beer, wine, NA, each on its own line. One overall figure of 24 percent can be hiding spirits at 31.

Once that is running every week, add the variance test. Work out what you should have used by taking the number of drinks your POS says you sold and multiplying by the ounces in each recipe. Then compare it to what your count says actually left the shelf. The difference between those two figures, worked out product by product, is your variance, and that is where over-pouring, waste, theft and unrecorded comps show up. One overall figure only tells you that a gap exists. Working it out product by product tells you which product it is, which shift it happened on, and who was behind the bar at the time.

Every Week

• Run the full bar count. Every week, without exception.
• Work out pour cost for each category separately, not as one overall number.
• Run the variance test and mark any product more than 3 percent out.
• Investigate every flag within 48 hours: pull its counts, check the shift records, confirm POS sales.
• Confirm opening and closing counts were done every shift.
• Close every investigation with a written finding, even an inconclusive one.
• Any product still unexplained moves to System 2 and gets treated as a loss.

The Paper This System Needs

Theft and Loss Prevention

A manager pulls the void report for the first time in eight months. One bartender is running four times the comp rate of everyone else, always Friday nights, always after 11pm, always on spirits, always when the floor is busy. She is not malicious. She is building regulars. But the bar is absorbing the cost and nobody was watching. Eight months of Friday nights.

Theft in bars is rarely dramatic. It looks like a void at 11:47pm on a busy Saturday. A comp that never got a signature. A delivery two cases short that nobody counted. Each one is small enough to rationalize. Together they are the difference between a bar that makes money and one that cannot figure out where it went. Your POS records theft. It does not catch it.

Run the math on one bartender voiding six 18 dollar transactions a shift, three shifts a week: about 16,000 a year, from one person, on one method, invisible without a report that breaks voids out by employee. The no-ring scheme is simpler and harder to catch: cash in, drink out, nothing rung, drawer comes up over, the overage gets pulled before the count. Giving away free drinks looks legitimate on the surface, and that is what makes it expensive. The bar pays for the drinks while the bartender builds a following of regulars who follow them to their next job.

The main value of a shift audit is not catching somebody in the act. It is that people work differently when they know an audit could happen at any time. If you always audit at 3pm on a Tuesday, you have trained your staff to be careful at 3pm on a Tuesday. Audit twice a week at times nobody can predict and everybody knows it happens but nobody knows when, and that changes how people work on every shift.

Run It Without Software

Almost all of this data already exists in your POS. The reason it catches nothing is that nobody runs the report and nobody ranks people against each other.

1. Find the exception report in your POS. Every system has one, usually called voids, comps, discounts, or employee exceptions. Set it to run by employee, for one week, and learn where it lives.
2. Build one sheet, seven columns. Employee, net sales, void count, void dollars, comp dollars, void percent of sales, comp percent of sales. One row per person, every week.
3. Rank, do not judge in isolation. Work out the average void percentage and comp percentage across all your staff, then look closely at anyone running more than double that average. Higher volume explains a lot of it: a server doing twice the sales should have roughly twice the voids. It does not explain a rate eight times higher than everyone else.
4. Reconcile every drawer before it leaves the floor. Counted by two people, signed, logged with the direction of the variance. A drawer that comes up over matters as much as one that comes up short. A drawer that is regularly over can mean drinks are being sold without being rung up, with the cash taken out later.
5. Run a shift audit twice a week at unpredictable times. Walk the bar mid-service. Check that every open ticket matches a drink on the rail, that the register is not sitting open, that comps have a name and a reason, and that jiggers are in use. Write down what you saw, every time, even when it is clean.
6. Count the delivery at the door. Short counts are the cheapest theft to prevent and the hardest to recover two days later.
7. Document before you act. One incident: document and watch. Two in 30 days on the same person: written corrective action. Confirmed cash theft: document fully and talk to your attorney before the conversation.

The Run List

• Daily: reconcile every drawer before it leaves the floor, counted by two people.
• Daily: log every void, comp, and no-sale by employee.
• Daily: confirm the delivery inspection was done on anything received.
• Weekly: rank every server on void and comp rate against the floor.
• Weekly: review drawer history and flag anyone with recurring same-direction variance.
• Weekly: run at least two shift audits, different shifts, different times, written up.
• Escalate on the written schedule, not on how the week felt.

The Paper This System Needs

Food Cost Control

A chef builds a beautiful menu. Creative, seasonal, priced against the competition. He never costs a single item. Eight months in, the owner pulls food cost: 42 percent. The menu is popular, the room is full most nights, and the restaurant loses money on every plate it sells. The food was not the problem. The math was, and the math was never done.

A chef who cannot tell you the cost of a dish to the penny is running your kitchen on feel, and feel runs 30,000 to 50,000 a year on 500,000 in food sales. The fix is three documents: a cost on every menu item, a portion standard posted at every station, and a waste log filled out every shift. That is the whole system.

The price you pay per pound and what that pound actually costs you once you have trimmed it are two different figures. A tenderloin bought at 18 dollars a pound where only 70 percent is usable after trimming really costs you 25.71 a pound. If you build your cost cards on the purchase price and ignore the trim, every protein dish on your menu is costed too low. Build your costs on the highest-volume items first: a 4 dollar over-cost on a bestseller running 200 covers a week is over 40,000 a year.

Over-portioning is not theft. It happens because cooks plate by eye, and what looks like the right amount varies from one cook to the next and from the start of a shift to the end of it. An average of an ounce and a half over on your best-selling protein at 200 covers can cost 40,000 a year on that one dish. Waste is the same story: a log that only records quantities tells you how much you threw away, while adding a reason to every entry, such as over-production, spoilage, prep error, a return or a quality problem, tells you why, and that is what lets you fix it.

Run It Without Software

1. List your top ten items by volume, not by price. Those ten are most of your food cost. Getting them right beats costing the whole menu badly.
2. Build a cost card for each one. Six columns: ingredient, purchase unit and price, yield percent, true cost per unit, quantity in the recipe, extended cost. Add the extended costs and you have the plate cost.
3. Always yield-adjust. True cost per pound is purchase price divided by yield percent. An 18 dollar pound at 70 percent yield is 25.71, not 18. Run a yield test once per protein: weigh it in, butcher it, weigh the usable portion, divide.
4. Set the price off the cost, not off the competitor. Plate cost divided by your target food cost percent gives the minimum price. If the menu price is below it, the item is a donation.
5. Post a portion spec at every station and put the tool next to it: scale, portioned scoop, ladle, ring mold. Pre-portion proteins during prep, not on the fly.
6. Audit two stations a week. Weigh five plates each against the spec, write the actual next to the spec, coach on the spot. Treat it as training, not discipline.
7. Log waste with a reason code every shift. Weekly, look at the top three categories by dollar and take one action on each with the kitchen manager.
8. Re-cost anything whose price moved. A supplier price rise makes a cost card wrong the day the delivery arrives, and cost every special before service rather than after.

Every Week

• Review the waste log: top three categories by dollar, one action each, with the kitchen manager.
• Run a portion audit on at least two stations: weigh five plates each against the posted spec.
• Update recipe costs for any ingredient whose price moved.
• Confirm reason codes are being used on every waste entry.
• Cost every special before service, not after.
• Check food cost for each category separately as soon as the overall figure is above target.

The Paper This System Needs

Vendor Control

An owner audits six months of invoices on a slow Tuesday with nothing else to do. She lays her orders against the invoices and checks line by line. By the time she finishes three vendors she has found 11,400 in overcharges. Not fraud. Price drift, substitutions billed at premium, and one delivery that came up a case short every week for four months. She had been signing without checking a single line.

Vendor overcharging is the quietest loss in the business. It does not spike and it does not flag in your POS. It accumulates one invoice at a time. A supplier will not offer you their best price if you never ask for it. That is not dishonesty, it is simply what any supplier does with a customer who never checks, and it costs you money for as long as you do not.

Short deliveries are the clearest example. Once the driver has left, you have no way to prove anything. You cannot call two days later and say you think a case was missing, because by then it is your word against their paperwork. The count you do at the door, before you sign, is the only thing that protects you. Telling the driver you think something is missing gets you nowhere. A written note on the invoice giving the invoice number, the date, what you ordered, what actually arrived and the dollar value of the difference is a credit request, and suppliers process credit requests.

Small drift compounds. A 1 litre spirit that creeps from 22.40 to 24.50 over a quarter at 18 bottles a week is a couple of hundred dollars on that one product alone, and you usually have twenty of them. The quarterly review is where you get it back: sit down with the rep with your variance history and a competitor price sheet, and ask for a match or an explanation. Show up with data and the conversation changes.

Run It Without Software

1. Build a price book. One sheet: product, supplier, pack size, current price, previous price, and the date it changed. Fill it from your last three months of invoices. This single sheet is what makes drift visible, and it takes an afternoon to build.
2. Have the order in your hand when the truck arrives. Not in the office. At the door. You cannot check what you did not write down.
3. Count every case before you sign. Short counts not caught at the door are not recovered. Open anything that should be sealed and looks wrong.
4. Check the invoice line by line against the order. Quantity, pack size, and unit price on every line. Flag anything more than 2 percent above your price book.
5. Write the discrepancy on the invoice before you sign it, with the invoice number, date, ordered quantity, received quantity, and the dollar gap. Photograph it. Give a copy to the driver.
6. Update the price book the same day for anything that moved, so next week's check has the new baseline.
7. Monthly, review your top spend items. Flag anything up more than 5 percent, or where another vendor is 8 percent cheaper.
8. Quarterly, sit down with each rep with the price book and competitor pricing. Ask for a match or an explanation. Confirm whatever you agree in writing after, not on the phone.

The Run List

• Every delivery: pull the order before the truck arrives. Know what you ordered and at what price.
• Every delivery: count every case before you sign.
• Every delivery: check the invoice against the order, flag anything over 2 percent above your price.
• File a discrepancy the moment you find one, while the driver is still there.
• Monthly: review top spend items. Flag anything up over 5 percent, or 8 percent cheaper elsewhere.
• Quarterly: sit down with each rep with your history and competitor pricing. Confirm terms in writing.

The Paper This System Needs

Prime Cost

An owner has been open three years. The room is full most nights, sales up 18 percent over last year. An accountant asks him his prime cost. He does not know what that means. She works it from six months of P&Ls: 71 percent. He has run a full bar and a busy kitchen for three years keeping less than 29 cents of every dollar. The room was full. The business was not working.

Prime cost is your cost of goods sold plus your labor, measured against your net sales. It is the one figure that covers the whole operation at once, and every other number in this playbook ends up in it: pour cost, food cost, what you save on vendors, and what you lose to theft. If prime cost is where it should be, the business is working. If it is too high, one of the first four systems is not doing its job, and you find that out now rather than when the P&L arrives a month later.

A monthly number tells you what happened. A weekly number gives you time to do something about it. A 5 point spike on 300,000 in monthly sales is 15,000, gone before you see it if you wait for the accountant. Catch it Monday of week three and you cap the damage at seven days instead of thirty.

Run It Without Software

1. Work out your cost of goods sold for the week. Opening inventory plus purchases minus closing inventory, food and beverage separately, then added together. This is why System 1 comes first: without a real count there is no real cost of goods sold.
2. Work out fully loaded labor. Gross wages plus employer payroll taxes plus benefits plus workers comp, plus any owner time on the floor valued at what you would pay someone else to do it.
3. Work out net sales. Gross sales minus comps and discounts, and excluding sales tax. Never register totals.
4. Prime cost = (cost of goods sold + labor) divided by net sales, times 100.
5. Compare it to your concept's band in the benchmark table above, not to somebody else's concept.
6. When it moves, run the split test. Did your cost of goods sold move more, or did your labor? Whichever moved more is the one to work on. If it was cost of goods sold, find which category. If it was labor, find which shifts or which department.
7. Act on the trend, not the week. One week with an explanation can wait. Three weeks going up in a row is a problem with how the business is set up, not a bad week.

Every Monday

• Read prime cost on last week's numbers.
• If it is above target, work out whether your cost of goods sold or your labor moved more.
• If it was cost of goods sold, find which category. If it was labor, find which shifts or department.
• Act this week. One week with an explanation can wait. Three weeks going up in a row is a problem with how the business is set up.
• Confirm net sales is after comps and discounts, never register totals.
• Confirm labor includes taxes, benefits, and owner time, not just wages.

Putting It In Place and Keeping It Running

An owner buys three cost control books over two years. She starts all three and finishes none. The first week of each feels productive. By week three service gets busy, a manager calls in sick, and the spreadsheet sits unopened. Three years later she still does not know her pour cost. The problem was never the system. What was missing was a plan in a set order, with a name against each task, a date on it, and one number to check that it is working. Motivation runs out after about ten days. A written routine does not, because it does not depend on how anyone feels that week.

Most efforts fail at the 45-day mark, not week one. A daily log stops getting filled. A Friday audit gets skipped twice. The invoice audit falls two weeks behind. None of it feels like failure, each feels like a one-time exception. A system that only happens when you personally make it happen is a habit rather than a system, and a habit does not survive a holiday, an illness, or a sudden busy stretch. What fixes that is writing the order of work down and putting a name and a day against each part of it.

The First Four Weeks

Week 3 feels uncomfortable the first time you tell experienced bartenders to use a jigger, or post specs in a kitchen that has run on feel for years. Do it anyway. Professional, direct, not apologetic. This is how the bar runs now.

Work the Rollout

Put a name and a day next to each of these before you start. A task with nobody’s name on it only gets done when somebody feels like doing it, and that is exactly the failure this system exists to prevent.

• Week 1: Run the first full count and price it.
• Week 1: Work out real pour cost, food cost, and prime cost. Change nothing.
• Week 2: Start opening and closing counts on every shift.
• Week 2: Start the daily void, comp, and no-sale log by employee.
• Week 2: Start drawer reconciliation before every drawer leaves the floor.
• Week 2: Start checking every delivery at the door, and count again at week end.
• Week 3: Sign the pour standards policy with every bartender.
• Week 3: Post portion specs at every station and put the tools next to them.
• Week 3: Hold the accountability talk with the whole team.
• Week 3: Build cost cards on your top ten items and run your first shift audit.
• Week 4: Run every system in the same week and read prime cost on Monday.
• Day 45: Check which system slipped first. It is always one of them. Restart it.

The Numbers Do Not Lie

A full-service bar in Cincinnati, about 1.1 million a year. The owner ran the whole system: weekly pour cost, a variance report every Monday, daily counts, void tracking, an invoice audit on every delivery, prime cost pulled before she looked at anything else. After 90 days she sat down with her accountant. Pour cost down 4.2 points. Food cost down 3.1. Two vendor disputes recovered 2,300 in the first six weeks. Prime cost 57 percent, under 60 for the first time in three years. She added no new revenue. She stopped losing the revenue she was already bringing in.

What separates bars that make money from bars that wonder where it went is not location, concept, or talent behind the bar. It is measurement and process. The tools are not complicated and the math is not advanced. What it takes is showing up Monday and pulling the number whether or not you feel like it. The week it feels least urgent is almost always the week something is quietly going wrong.

The gap between what you are making and what you should be making is not a mystery. It is a measurement problem. Start measuring tonight.