Pub stock loss: pouring, unrecorded waste, or a measuring error? How to tell from the numbers

A stocktake tells you how much is missing. It doesn't tell you why, and the why is the only thing you can act on. In my experience the missing pounds come from one of three places: pouring (over-measures, froth, top-ups, the free splash), waste that happened but never got written down, or a measuring error, where the stock was never missing at all and the count was wrong. Each one leaves a different fingerprint in the numbers. This is how to read them apart before you change anything, and definitely before you have a difficult conversation with anyone.

I run a wet-led pub in Washington, Tyne and Wear. I've had five liquor audits in eight months, one of them £756 down, and the useful thing I learned from arguing about that one is that the total is nearly meaningless. The pattern underneath it is everything.

Start with the shape, not the total

Before you look at the bottom line, look at how the loss is spread. Get your variance per line, in pounds and in units, for at least two consecutive counts. One count can't tell you anything about cause. Two can. Three is better.

Then ask four questions of the spread. Is the loss concentrated in a few lines or smeared across everything? Is it proportional to volume, so the fastest lines lose the most? Does the same line lose every count, or does it lose this time and gain next time? And does it follow a calendar, spiking on a specific weekend or after a specific event? The answers point at the cause more reliably than any gut feeling.

Fingerprint one: pouring

Pouring loss is over-delivery. It's the 25ml measure that's really 28 because the optic's tired, the head that gets topped up on request, the double that's a bit generous for a regular, the cocktail spec nobody follows, and on draught, the froth. It is almost never one person and it is almost never deliberate. It's habit, and habit is proportional to volume.

So the fingerprint is: the loss is smeared across the busy lines, it scales with how much each line sells, it's roughly consistent from count to count, and it's expressed as a fairly steady percentage rather than a fixed number of bottles. House vodka and the main lager lose the most in pounds because they sell the most. Slow lines look fine.

Draught has its own wrinkle. British Beer and Pub Association guidance says a pint served with a head only has to be 95% liquid, so up to 5% of every pint is legally froth that never rings through the till as volume. A naive stock system reports that as missing. If your draught lines are all showing a loss of a few percent and your spirits aren't, that's not pouring, that's the head allowance and your tool not understanding British pubs. The calibration guide covers how to separate the two.

What you do about real pouring loss is boring and works: check the optics, standardise the measure on the busy spirits, agree the head, and re-count in a week. I wrote up what staff pours actually cost in the staff pours guide, and it's a training problem, not a trust problem.

Fingerprint two: unrecorded waste

Waste is stock that really did leave the building without being sold: the cask that went off, the line clean, the bottle that got knocked over, the returned pint, the ullage you drew off, the wine that oxidised, the kitchen borrowing a bottle of brandy. All legitimate. All invisible to a stocktake unless somebody wrote it down at the time.

The fingerprint is lumpy. Waste loss is concentrated in specific lines, in specific counts. One cask ale shows an 18-pint hole and the other three are fine. The white wine by the glass is 2 bottles down after a quiet week. The loss doesn't scale with volume, it doesn't repeat evenly, and when you ask around, somebody usually remembers. "Oh, that one went cloudy on Sunday." If you keep a wastage book and the losses line up with the entries, it's waste. If you don't keep one, waste and pouring look identical on paper and you'll blame the wrong thing.

Draught deserves its own line here. A cask that's on for four days, a keg that gets changed mid-session with a foaming line, a line clean that pulls through two pints per line and nobody logs it, these add up to a persistent draught hole that looks like a pouring problem and isn't. If the hole is on draught only and it's biggest on the lines that get cleaned or changed most, look at the cellar before the bar.

The fix is a wastage log people will actually use, which means one that takes ten seconds. Line, amount, reason, initials. Then the next count nets it off and what's left is the real question.

Fingerprint three: measuring error

This is the one nobody wants to hear, and I'd bet it's the most common. The stock isn't missing. The count is wrong. The bottle was called at three tenths when it was four. The dip was read on a keg that was tilted. A case was counted twice. A delivery went in on the wrong date, so it landed in the wrong period. The sell price was put in inc-VAT against a cost price that's ex-VAT, so the whole GP% is seven points flattered. Or the opening stock was itself an estimate, so every count since inherits its error.

The fingerprint is a flip-flop. A line that's 6 units down this count is 5 units up the next. The total swings but the two-count average is near zero. Spirits that get eyeballed swing; packaged lines that get counted by the case don't. Part containers swing; full ones don't. If your variance report looks like a heart monitor rather than a slope, you don't have a loss, you have a counting method.

Two more tells. If the loss appeared the week you changed something (a new product line, a new price, a new person counting), it's probably the change, not the stock. And if the loss is in value but not in units, or vice versa, it's a price or a VAT setting. Ex-VAT versus inc-VAT catches more people than anything else on this list.

The fix is to take the guessing out of the count. Weigh the open spirits instead of eyeballing them, calibrate the dip on every container size you use, count the same way every time, and check the opening stock once properly instead of inheriting it. Tighten the counting method first and a surprising amount of the "loss" goes away before anyone has changed how they pour.

The diagnosis table

What about theft?

People ask. Here's my honest position after nineteen years behind the pumps: it's the last thing to consider, not the first, because the three causes above explain nearly everything and are fixable without accusing anyone. A genuine problem tends to look like concentrated, repeated loss on specific high-value lines that doesn't move when you fix the count and log the waste, and it's usually paired with till patterns, voids and no-sales, rather than stock alone. If you get there, you get there with two months of clean numbers and a wastage book, not a hunch after one bad audit. I'd rather find out it was the dip stick before I've said something I can't take back.

The order to work in

Fix the count first, because until the measuring is tight you can't trust the shape. Weigh the spirits, calibrate the containers, sort the VAT setting, count the same way twice a fortnight apart. Then start the wastage log and give it a month. Whatever loss is still there, smeared across the busy lines and steady, is pouring, and pouring is a training conversation you can have with a straight face because you've done the work.

That's the whole method. The acceptable variance guide tells you when the number left over is normal. This one tells you where it came from.