You're counting bottles. You sell ounces.
Your inventory says zero of the good rye. You know there's most of an open bottle behind the bar, and you know it moves about twice a month, so you don't order. You just corrected your own inventory system in your head. You do that every week, on a dozen items, and you've stopped noticing you do it.
The wrong unit
A bottle is a container, not an amount
Almost every inventory system counts containers. Bottles, cans, kegs. Not because anyone decided that was the right unit, but because a container is the thing a system can see: it arrives on an invoice as one line, it gets scanned as one object, it leaves as one empty.
Liquid doesn't behave like that. A 750ml bottle is about twenty-five ounces, and it leaves in whatever size the drink called for. From the day it opens to the day it empties, it is part full. It stays part full for weeks, and that is the state it is in almost all of the time.
The container count is accurate on the day the bottle arrives and the day it runs out. Every day in between it is telling you about a box, and you are asking about a drink.
What that costs
The number stops being a number and becomes a suggestion
The damage isn't that the count is wrong. It's that the count is wrong in a way you have to fix yourself.
A slow-moving top shelf item sits part-full for a long time. The system shows zero full bottles and puts it on your order sheet. You know better, so you take it off. Next week it's back. The system has no way to learn that you were right, so it will suggest it again, and you will override it again.
That overriding is real work, and it lives in one person's head. It doesn't survive that person taking a week off. It doesn't transfer to the new manager. And it teaches everyone that the inventory number is a starting point rather than a fact. A number you have to mentally correct isn't inventory. It's a suggestion.
Meanwhile the reverse costs you too. Something that looks like plenty on a container count can be three ounces from empty, and nobody finds out until a guest orders it.
The right unit
Count what's in the bottle
So we count the liquid. An open bottle isn't a zero and it isn't a one. It's an amount, and the count carries that amount. Everything that draws on the bottle comes off that one number, whatever the drink was and whatever size it was poured at.
That sounds like a small change. It isn't, because it changes what a count can answer. A container count can only tell you what you own. An amount can be compared to what you actually pour, and once those two things are in the same unit, the count can be subtracted from.
What that unlocks
How long it lasts, not how much you have
Once a count is in ounces, "how much do I have" turns into "how long will it last."
It has to be ounces and not pours. A pour is a container too, just a smaller one, and counting in pours quietly assumes every drink took the same amount. Halves, doubles and cocktails all draw on the same bottle at different sizes. Volume is the only unit that adds up.
Work it yourself. Twenty-five ounces in a full bottle. Add up everything that drew on that gin last week, the singles, the halves, the doubles, the ounce that went into every Negroni, and say it came to eighteen ounces. A full bottle is nine or ten days. The one behind the bar that's a bit over half gone is under five. That is a different sentence than "zero full bottles," and it's the sentence an owner is actually trying to arrive at when they look at an inventory report.
It also means the thing the manager was doing in their head is now written down. Not because we out-thought them. They were right, and the system had no way to record it.
Nothing gets ordered on your behalf. The count says how many days you're carrying. What you do about that is your call, the same as it was before.
Where this doesn't reach
Somebody still has to look at the bottle. How much is left in an open one is a physical fact, and no system knows it until a person checks.
The projection is only as good as the history behind it. A brand new bar has none. A brand new item has none. And a week that doesn't look like the last ten weeks, a wedding party, a heat wave, a long weekend, will break the estimate. It's an average of what already happened, not a forecast of what's next.
Days of cover is a better question than bottles on hand. It is still an estimate, and we would rather say that than let you plan around it as though it were a promise.