The two numbers a bottle has
Every container has a paper yield and a real yield, and the gap between them is where bar margin goes. The paper yield is volume divided by the spec pour. A 750 ml bottle is 25.36 fluid ounces, so at an ounce and a half it is 16.9 pours. The real yield is what the register recorded against that bottle, and it is reliably lower.
Nothing about that is scandalous. The gap is made of ordinary things: the pour that overshoots, the drink remade because it was wrong, the round comped for a regular, the shift drink, the bottle knocked over, the last half ounce that will not come out cleanly, the cocktail whose recipe uses more than the field says it does. The loss allowance field is where you put all of it, and the useful version of that number is measured rather than assumed. Count opened containers over a fortnight, pull the pours the register rang for the same products, and the difference is your allowance.
Because the container cost is fixed, this calculator spreads it across the whole drinks it actually yields rather than the fractional paper figure. That keeps the arithmetic honest: the drinks a bottle sells, multiplied by the cost this page assigns to each, add back to what the bottle cost, and the reconciliation line shows the rounding.
What a quarter ounce costs
An extra quarter ounce on a one-and-a-half ounce pour is a 16.7 percent increase in the product going into every glass, and it is close to invisible in a rocks glass with ice. Run the numbers on a bottle that yields 15 drinks: at 1.75 ounces it yields 13. Two drinks a bottle is 13 percent of the revenue that bottle was supposed to produce, and it compounds across every bottle on the well.
The annual figure the calculator produces is deliberately blunt. At 140 drinks a week and a quarter ounce over, that is 1,820 extra fluid ounces a year from one product, which is roughly 72 bottles of 750 ml poured away, and at an $11 menu price it is over thirteen thousand dollars of drinks that never rang. Whether that number is real depends entirely on whether the overpour is real, which is why it is a field you fill in rather than an assumption.
The fix is not usually discipline. Jiggers, measured pourers and speed pourers with a counted rhythm all work, and all of them are cheaper than the loss. Free-pouring can be accurate, but only from people who train on a scale and get re-checked, and it degrades quietly when service gets busy.
Kegs, cocktails and things that are not one bottle into one glass
A keg is the same calculation with a bigger number in the size field. Set the unit to gallons, put in the keg volume your supplier states, and the loss allowance carries the foam, the line cleaning and the first few pints of a fresh keg. Draught loss is typically much larger than bottle loss and much more variable, since it depends on line length, temperature, gas pressure and how often the system is cleaned, so measuring it matters more here than anywhere else.
A cocktail with several spirits is several passes through this page. Run each spirit at its own pour size and container cost, add the resulting per-drink costs together, then add the modifiers, juice, garnish and ice dilution as ingredients in the recipe cost calculator. Fresh juice needs a yield of its own, because citrus is bought by the case and used by the ounce, and the portion yield calculator handles that conversion.
Pour cost against food cost
Beverage generally runs at a lower cost percentage than food and carries far less labour, which is why a bar attached to a dining room changes the whole economics of the room. That is also why the two are usually tracked separately: folding them into one figure hides a shift in the sales mix that is doing more to your margin than anything in the kitchen. If you want the two side by side, the menu pricing calculator handles the food half by dish and the cafe and restaurant profit calculator puts both into a monthly picture.
Beverage also has an inventory problem that food does not, because a bottle sits open for weeks and its remaining contents are countable to a tenth. That makes bar variance measurable in a way kitchen variance is not, and a weekly count against register pours is the single most useful control most bars are not doing. Purchasing rhythm and safety stock are ordinary inventory arithmetic, which the inventory order calculator covers.
What this page will not tell you
It says nothing about how much alcohol is in a drink, what a standard serving is, or anything to do with responsible service. Serving measures are regulated in many jurisdictions, the definition of a standard drink varies between countries, and licensing conditions are set locally and enforced locally. Those obligations sit with your licensing authority and your own training programme, and none of them can be worked out from a cost figure. Treat this strictly as a costing tool.
Questions people ask
How many drinks should a 750 ml bottle give?
On paper, volume divided by the pour: 25.36 fluid ounces at 1.5 ounces is 16.9, at 1.25 ounces is 20.3, at 2 ounces is 12.7. In practice you will not get the fractional pour and you will not get everything out of the bottle, so the honest planning figure is the whole number after your own loss allowance. Whether your real yield is 16 or 13 is not something anyone can tell you from outside, and it is worth measuring rather than assuming, because that single number sets your entire beverage cost. The measurement is a count of opened bottles against register pours over a couple of weeks.
What should my pour cost percentage be?
Beverage cost percentages get quoted as ranges and the ranges are close to useless without context, because they depend on the split between spirits, wine, beer and cocktails, on your price point, on your region, and on whether draught and bottled beer are in the same figure. Spirits, wine and beer usually run at quite different percentages within the same bar, so a single blended number can move several points purely from a change in what people ordered. The number that actually tells you something is your own, tracked by category, month over month. A category that drifts is a signal worth chasing, and the level it sits at matters much less than whether it moved.
Should the loss allowance include comps and staff drinks?
It depends what you want the number to mean. Putting them in gives you a true cost per sellable drink, which is the right figure for pricing, since those pours are real product that produced no revenue. Leaving them out and tracking them separately is better for control, because comps are a management decision and spill is an operational problem, and mixing them means a rising number does not tell you which one moved. A reasonable compromise is to keep spill, breakage and residue in the allowance here, track comps and staff drinks as their own line, and be aware that your pricing needs to cover both.
Does this work for wine by the glass?
Yes, and wine is where the fractional yield hurts most. A 750 ml bottle at a five-ounce pour is 5.07 glasses on paper and five in reality, so the entire remainder is loss before you have spilled anything. At a six-ounce pour it is 4.23, which rounds down to four and throws away a quarter of the bottle in the rounding alone. That is why glass pours cluster at sizes that divide the bottle cleanly. Wine also has an open-bottle problem that spirits do not: an opened bottle that does not sell out has a limited useful life, so slow-moving by-the-glass listings carry a waste cost this calculator handles only through the loss allowance.
Why does the calculator spread the bottle cost over whole drinks instead of the exact fraction?
Because the bottle costs what it costs regardless of how the last two ounces work out, and money that is not assigned to a drink is money that quietly vanishes from the costing. Dividing by the fractional yield produces a slightly lower cost per drink that never reconciles against the invoice. Dividing by the whole drinks the bottle really produces means the drinks sold, times the cost per drink, add back to what you paid, which is the reconciliation line in the output. The difference is small on a bottle that yields sixteen and quite large on one that yields four, which is exactly the case where you want the arithmetic to be strict.