Pour cost and bar program management: where beverage margin disappears

Beverage is the highest-margin thing you sell and the easiest to give away. A quarter-ounce of extra gin per drink costs nothing visible — no waste bin, no comp ticket, no line on a report. It just shows up weeks later as a pour cost that's four points higher than it should be and nobody can explain. Food waste you can see in a trash can. Bar waste evaporates.

The upside: beverage inventory is a small number of high-value, countable, non-perishable SKUs, which means bar variance is knowable to the ounce in a way food variance never is. If you're willing to count, the bar will tell you exactly where the money went.

Pour cost, and what "good" looks like

Same shape as food cost, applied to the beverage side:

Pour cost = cost of beverage sold ÷ beverage sales

And the same trap: cost of beverage sold comes from inventory (beginning + purchases − ending), not from what you bought this month. Divide purchases by sales and you're measuring your ordering, not your pouring.

Commonly cited benchmarks by category:

(These are industry benchmarks from bar-inventory and POS vendors rather than an audited survey — consistent across independent sources, but treat them as guideposts.)

Two things to take from that table. First, spirits carry the best margin and wine the worst, which means your beverage mix moves your blended cost as much as your pricing does — a wine-led program running 30% can still be excellent because gross profit dollars per bottle are large. Second, and more important: a single blended number hides everything. Track pour cost by category or you'll never find the problem.

The bottle math every bar should know cold

• A 750ml bottle is 25.4 fluid ounces; a liter is 33.8.
• At a 1.5 oz standard pour, a 750ml yields 16.9 pours — most operators plan on 16 to allow for spillage.
• Cost per ounce = bottle cost ÷ ounces. A $30 bottle is $1.18/oz, so a 1.5 oz pour costs about $1.77.
• Drink cost = spirit + mixers + garnish, summed, then divided by menu price to get that drink's pour cost.

Now the arithmetic that should end the free-pour debate. Pour 1.75 oz where the recipe says 1.5 — a barely perceptible overpour — and you're giving away 0.25 oz per drink. Do it a hundred times and you've poured away 25 ounces: one entire 750ml bottle, gone, with no transaction, no waste log, and no explanation. A busy bar does that several times a week without anyone doing anything wrong on purpose.

That's why jiggers and measured pour spouts exist, and why craft programs treat jiggering as standard rather than as a slight on the bartender's skill.

Variance: making the loss attributable

Theoretical pour cost is what your POS says should have been used, given the drinks sold at recipe spec. Actual is what the count says was used. The gap is your variance, and unlike in the kitchen — where trim and yield make theoretical cost inherently fuzzy — a bottle is a bottle. The variance is real and it points somewhere.

Where it usually points: over-pouring, spillage, unrecorded comps and staff drinks, breakage, bad kegs, oxidized wine by the glass, and theft. I'm deliberately not quoting you an industry shrinkage percentage, because every figure in circulation traces back to companies selling bar-inventory software and none of them cite real data. Run your own variance — your number is the only one that means anything.

What does the arithmetic say about stakes? At a ~20% pour cost, every dollar of product lost represents about five dollars of revenue you never collected. That multiplier is why bar shrinkage punches so far above its apparent size.

The controls that actually work

1. Measured pours. Jiggers or measured spouts, consistently. This is the single highest-impact control.
2. Weekly counts on a consistent cadence — same people, same units, same time relative to deliveries — so variance lands inside a period short enough to attribute to a shift.
3. POS-to-inventory variance reporting. This is what turns "pour cost feels high" into a specific bottle on a specific night.
4. Per-bartender reporting and spot checks, plus real authorization controls on comps and voids.
5. Time-based pricing as a POS price rule, not a manual discount — otherwise happy hour shows up as untracked comps and corrupts both your revenue data and your variance.

Counting discipline is the same muscle as the inventory playbook; the bar is just the part of the building where it pays fastest.

The POS features that protect bar margin

Two, specifically:

Tab pre-authorization. A hold placed when the tab opens verifies the card and stores it for close-out, and EMV pre-auth can pull the cardholder name to auto-label the tab. It's the main defense against walkouts. One honest limitation: pre-auth holds typically last around 36 hours and don't guarantee funds for the full final check — a guest can run past the authorized amount, so it reduces walkout losses rather than eliminating them.

Per-bartender sales and variance reporting, so shrinkage has a name attached instead of being a monthly aggregate mystery.

Both are why bar-specific POS capability matters more than a generic feature list — the argument I made concept by concept in the business-type guide, and in more depth on the bar POS page.

Disclosure: I work at Katalyst, and pre-auth tabs and per-bartender variance are things we build. But the highest-return item on this list costs about four dollars: a jigger. Measure the pour, count weekly, track by category, and let the variance tell you where the money went. The bar is the most controllable margin in the building — it just refuses to be controlled by feel.

Related Katalyst products

• See Katalyst analytics