Pub stocktaking mistakes: the twelve that quietly wreck your result

Most bad stocktake results aren't theft. A lot of them aren't even loss. They're mistakes in the count itself, and the nasty thing about a counting mistake is that it looks exactly like missing stock. I've had five liquor audits in eight months, and the education wasn't the audits, it was finding out how many ways my own numbers could lie to me. Here are the twelve I'd check before believing any result, mine included.

They fall into four piles: mistakes in the counting, mistakes in the valuing, mistakes in the timing, and mistakes in what never got written down between counts. The last pile is the biggest, and it's the one nobody blames because it doesn't happen on count day.

Counting mistakes

1. Guessing part bottles by eye

Squinting at a back bar of forty open bottles and calling each one "about a third" puts a couple of GP points of pure noise into the result before anything real has happened. That's my own working estimate from doing it both ways, not a published figure, but the direction is not in doubt: a tenth guessed wrong on a £26 litre of vodka is £2.60, and you're making forty of those guesses. Weighing removes the guess entirely, spirits sit around 0.95 g/ml, and the part-bottles guide has the method and the arithmetic.

2. Wrong sizes in the library

When I imported my own 270-line product list, a batch of wines went in at 700ml that are 750ml bottles. Every full bottle then counted 50ml light, which reads as loss that never happened. Kegs are worse: a 30-litre keg set up as a 50 shows half your lager missing. Boring to check, expensive to skip, and the setup guide shows where these errors creep in.

3. Missing whole locations

The function room shelf. The kitchen's cooking wine. The crate of mixers behind the cellar door that came off the last dray. If it was paid for and it isn't counted, it's a shortage on paper. Walk the whole building with the count sheet once and write down every place stock hides, then count the same places every time.

4. Counting kegs at their theoretical yield

A 50-litre keg is 88 pints on paper. By the time lines are cleaned, the keg's changed and the fobbed bits have gone down the drain, mid-80s is a good real-world return. Judge the bar against 88 and it will look short every single week while nothing is wrong. The draught wastage guide covers what normal actually looks like.

Valuing mistakes

5. Mixing VAT bases

The same pint is a 64.9% GP measured ex-VAT and 70.8% measured VAT-inclusive. Compare a stocktake done on one basis against a target set on the other and you'll spend a week hunting five points of GP that live in the tax, not the cellar. One basis, everywhere, forever; the GP report guide shows the four different answers the same pint can give, and the ex-VAT guide explains which to use where.

6. Valuing stock at selling price

Your stock holding is what you paid for it, not what you'll sell it for. Value the cellar at retail and you'll carry a number roughly three times too big, and your accountant will unpick it at year end anyway, because the rule for accounts is cost or net realisable value, whichever is lower. The year-end valuation guide has the actual HMRC references.

Timing mistakes

7. Deliveries straddling the count

A keg delivery that arrives Thursday but gets booked in Friday, with the count in between, appears as a full keg of surplus in one period and a full keg of loss in the next. As an illustration, one £160 keg booked into the wrong period moves a £5,000 ex-VAT week by about three GP points on its own. Book everything in before you count, or leave it sealed, off the count sheet and off the books until after. The delivery-delays guide and the short-deliveries guide cover both ends of this.

8. A till period that doesn't match the count period

Count Sunday morning, then pull a sales report that runs to Saturday close, and Sunday lunchtime's sales are in your stock movement but not your revenue. The result reads as a surplus that isn't there, and next period a shortage that isn't either. The sales report and the count must start and end at the same minute. It's a thirty-second check and it catches more "weird results" than almost anything else on this page.

Recording mistakes, the between-counts pile

9. Line cleaning never logged

Vianet's dispense data, gathered across roughly 11,000 pubs, found 27% of pints being poured through lines outside a seven-day clean. Cleaning done properly throws real pints down the drain every week, a pint and a quarter per ten metres of standard python per line, and if nobody writes that down, count day files it under unexplained. The line-cleaning guide has the per-line arithmetic and the wastage log guide the recording habit.

10. Comps and staff drinks off the books

A pint on the house costs you about £1.56 in stock and shows up in the variance at its £5.34 selling price if it was never rung through. One unrecorded comp a night is nearly £2,000 a year of "missing" stock that was actually hospitality. Zero-priced till button, ten seconds, done; the comps guide has the sums.

11. Prepaid pints left in the pump

"Leave one in the pump for Tommy" is a sale this week and a pour next week. Ring twelve of them before count day and you show a surplus; pour them after and you show a shortage, and both send you hunting for a problem that's actually a kindness with no system behind it. The prepaid pints guide shows the systems that reduce it to one count-day number.

Reading mistakes

12. Treating one count as a verdict

A single result is a photograph, and photographs blur. The mistakes above can all land in one period and unwind in the next, which is why a shocking number followed by a glowing one usually means the counting, not the bar, changed. Watch the trend across a couple of counts before acting on anything, know what an acceptable variance looks like, and if a real loss survives a careful second count, work it methodically: the unexplained loss guide is the plan. What I'd never do again is start with the staff. Every time I've seen that done, the real cause was on this page, and the good people had already started looking for other jobs.

The checklist

Print this, pin it inside the cellar door. Worked figures above are illustrative author arithmetic unless a source is named.

Mistake	What it does to the number	Fix
Guessed part bottles	±2 points of noise	Weigh them
Wrong sizes in library	Phantom loss or surplus	Audit sizes once
Missed locations	Paper shortage	Fixed count route
Theoretical keg yields	Permanent "short" reading	Judge against real yield
Mixed VAT bases	~6 GP points of confusion	One basis everywhere
Stock valued at retail	Holding ~3x overstated	Value at cost
Straddling deliveries	Surplus then shortage	Book in before counting
Till period mismatch	False swing both ways	Same start and end minute
Unlogged line cleaning	Weekly pints as "loss"	Log every clean
Unlogged comps	£5.34 loss per £1.56 pint	Zero-priced button
Prepaid pints	Swing both directions	One recording system
One count as verdict	Wrong accusations	Trend over two counts

Common questions

What's an acceptable stocktake variance for a pub?

Once the mistakes on this page are dealt with, a well-run wet bar can hold variance under 1% of sales, and plenty run tighter. The honest answer is that the acceptable number is the one that's explained; the variance guide goes deeper.

How often should I count to catch these?

A full count monthly and a five-line spot check weekly is the shape that catches most of this within days instead of quarters. The frequency guide covers how that changes with your size and your worry level.

Do I need software to avoid these mistakes?

No, and I'd be lying if I said otherwise. A disciplined spreadsheet with fixed routes, one VAT basis and booked-in deliveries avoids about half of this list. What software closes is the recording pile, comps, wastage and cleaning logged in seconds at the bar rather than remembered at midnight, plus the size checks and the weighing. The spreadsheet guide is the fair comparison.

Which mistake should I fix first?

The recording pile, mistakes nine to eleven, because those keep polluting every future count until the buttons and the log exist. Then the library sizes, because that's a one-off job that pays forever. The counting-technique fixes come naturally after that.