On this page
Key takeaways
- A pour cost report does not tell you who, what, when or why: it averages away the person, the product, the shift and the cause.
- Pour cost is a legitimate benchmark and a good health check, but a poor diagnosis.
- The same headline number can come from a whole team pouring slightly heavy or from two people well over spec.
- Because it comes from a physical count, pour cost arrives weekly at best and usually monthly, long after the shift.
- A finding only changes something when it ends in a named person, a specific action and a date.
What is a pour cost report good for?
Pour cost, the cost of the liquor sold divided by the revenue it produced, is a legitimate and useful benchmark. It tells you roughly whether your beverage program is in the right shape, it is comparable between periods, and it is the number your accountant and your investors will ask for.
None of that is in dispute. The problem is what happens when an operator tries to use it as a diagnostic instead of a benchmark.
Four things the number has averaged away
Who
A bar with a twenty two percent pour cost might have a team that is uniformly a little heavy handed, or eight people at spec and two people well over. Those are completely different situations requiring completely different responses, and they produce the same headline number.
What
Aggregate pour cost hides which products are actually leaking. A well spirit running heavy and a premium spirit running heavy have very different dollar consequences per ounce, and the blended figure treats them identically.
When
Variance concentrated on Friday and Saturday late is a different problem from variance spread evenly across the week. The first suggests pressure, staffing or supervision; the second suggests spec, training or a measurement issue. The monthly number cannot distinguish them.
Why
This is the big one. Pour cost is silent on cause. Overpouring, unentered comps, a short delivery and unlogged breakage all move the same number in the same direction, and each needs a different fix and a different owner. We worked through how to tell them apart in calculating bar variance.
The timing problem on top
Pour cost is usually calculated on a counting cycle, which means it arrives weekly at best and monthly in most operations. So even where the number does tell you something real, it tells you weeks after the behaviour that caused it, when the shift is unrecoverable and the conversation is about something nobody clearly remembers.
That is not a criticism of any particular tool. It is structural: a figure derived from a physical count cannot arrive faster than the counts.
How should you use pour cost, and what should you use instead?
Use pour cost as it is meant to be used: a periodic health check, and a trigger for a look rather than a conclusion.
For the look itself, the useful source is the transaction stream rather than the count, because that is where the who, what and when survive intact. Void and comp rates per person against the team, product mix by shift, and patterns that recur across weeks are all available continuously without anybody touching a bottle.
The count is still worth doing. It just works better as confirmation of something you already suspect than as the mechanism by which you discover it.
Making it end in something
Whichever way you find it, a finding that ends in a revised number has not changed anything. It has to end in a named person, a specific action and a date, or the same figure reappears next period looking like a fresh problem. That is the argument in the data to action gap.
Frequently asked questions
Is pour cost still worth tracking?
Yes. It tells you roughly whether your beverage program is in the right shape, it is comparable between periods, and it is the number your accountant and investors will ask for. Use it as a periodic health check and a trigger for a closer look, not as a conclusion.
Why can't my pour cost tell me what is actually wrong?
Pour cost is silent on cause. Overpouring, unentered comps, a short delivery and unlogged breakage all move the same number in the same direction, and each needs a different fix and a different owner.
What should I look at instead to find the problem?
The transaction stream, because that is where the who, what and when survive intact. Void and comp rates per person against the team, product mix by shift, and patterns that recur across weeks are available continuously without anybody touching a bottle.
Should I stop doing stock counts?
No. The count is still worth doing. It works better as confirmation of something you already suspect than as the way you discover it.
Does it matter which night the variance happens on?
Yes. Variance concentrated on Friday and Saturday late suggests pressure, staffing or supervision, while variance spread evenly across the week suggests spec, training or a measurement issue. The monthly number cannot tell them apart.