I spent a long time reading the bottom line of a stocktake and nothing else. Net figure, is it green or red, close the folder. It took five audits before I looked at the page above it, and the page above it was telling me something completely different to the page I’d been reading.

Five stocktakes told me the same thing five times

Here’s what my categories did across five Venners audits between December 2025 and July 2026.

Draught Lager: +£3,697, positive in four of the five, once at +£1,644 on its own. Draught Stout: +£870, positive in all five. Draught Ale: −£773, negative in all five. Minerals: −£1,111, negative in all five. Wines: −£962, negative in all five. Postmix −£1,222, Spirits −£643, Packaged −£589.

Read that again with one thing in mind. Genuine random variance flips sign. Some months you’re up, some months you’re down, because a count is a bit rough, a keg gives a bit more or less, someone’s heavy-handed on a Saturday. Five out of five in the same direction is not random. Five out of five in the same direction is a system doing exactly what it was told to do.

The net figure hides nearly all of it

Across those five audits, £11,598 of stock moved between categories. The net of all that was −£434. That’s roughly twenty-seven times as much movement as the headline number ever showed me.

The clearest example was 21 July. Net result: £12.84. Underneath it, £1,165 of stock had moved in opposite directions. On paper that was one of the tidiest stocktakes I’ve ever had. In reality it was £1,165 of things going wrong that happened to cancel each other out to within about a tenner.

Same story with gas. One visit carried £636 of gas adjustments — mixed gas at −£266.78 and CO2 at −£369.34 — which never appear anywhere near the headline figure. If you only read the bottom line, that visit looked fine.

A net of zero doesn’t mean nothing happened. It means two big things happened and they were pointing at each other.

Direction is the tell, not size

This is the part I’d want somebody to have told me two years ago. Don’t chase the biggest number. Chase the number that goes the same way every single time.

A one-off −£400 on wines is probably a count, a delivery on the wrong side of a date, or a bad week. Wines negative five audits running is a rule. Something in your setup is systematically converting wine into not-wine, and it will keep doing it every week whether anybody is looking or not.

Same with Stout positive five times out of five. A surplus sounds like good news. It isn’t. A consistent surplus means you’re selling stout that the till doesn’t know you sold, which means the money for it went into some other category’s takings, which means somewhere else on the report there’s a matching hole. It’s the same error viewed from the other end.

Lager +£3,697 and Ale −£773. Stout +£870 and Ale −£773. Those aren’t three separate problems. That’s one problem with three symptoms.

Hobgoblin and Hobgoblin Gold sit next to each other on my till

Which is, as far as I can tell, most of my variance.

They are next to each other because that’s how anyone lays out a till: same brand together, same category together, alphabetical. It makes complete sense right up until a Friday night when the bar’s four deep and the difference between the right button and the wrong button is about eleven millimetres and one word.

Pour the Gold, ring the standard, and the stocktake sees stock gone from one line with no sale against it and a sale against another line with no stock gone. Ale goes negative. Something adjacent goes positive. Do that a dozen times a week for a year and you get numbers that look like a serious problem and are in fact a layout problem.

Minerals down £1,111 and postmix down £1,222, both consistently, is the same thing in a different corner of the bar. Nobody on earth reliably distinguishes between six similar-looking soft drink buttons at speed, and half of them get rung as whatever the customer said rather than what actually came out of the gun.

How to check your own till buttons in an hour

This is dull and it works.

Print your product list off the till and read it as a stranger would. Anything where two lines share the first word, or the first eight characters, or look identical on a small screen, is a candidate. Write them down.

Check the selling price on every one of those pairs. If the two products are the same price, the till error costs you nothing in cash and shows up purely as a category variance — annoying but harmless. If they’re different prices, you’re losing actual money on every mis-ring, and that’s your priority list.

Look at your report by category, over at least three stocktakes, not one. You are looking for direction, not size. Anything that has gone the same way three times or more goes on the list.

Pair the positives with the negatives. For every category that’s consistently up, find the one that’s consistently down. Nine times out of ten they’re physically near each other, either on the till screen or on the bar.

Then move the buttons. Different colour, different row, different screen, whatever your system allows. Put a gap between them. I moved two buttons apart and it did more for me than three months of talking about tightening up.

Check your yields separately. Venners’ 2021 Yield Report, drawn from around 22,000 stocktakes across more than 3,000 sites, found packaged beers and ciders averaging 97.6% yield against an industry expectation of 99.5%, and put up to £10,000 per site per year as recoverable. Wine came out best at 98.84%. Packaged should be the easiest thing in the building to get right — it comes in a bottle, you sell the bottle — so if packaged is off, that’s a counting or a ringing problem, not a pouring one.

This exonerates your staff, it doesn’t blame them

I got this wrong for a while and I’d rather say so than pretend otherwise. My first instinct at a bad result was to look at the rota for the period and wonder who was on.

Venners’ observation audits, reported in the Morning Advertiser in December 2019, found losses at 100% of the sites they audited, and attributed 46% of loss to human error against 39% to theft. Error is the bigger share. And “human error” in practice mostly means an honest person pressing a button that is nine millimetres from the right one while somebody waves a card at them.

Once you see it as a layout problem, you can fix it. You can move a button on a Tuesday afternoon. You cannot fix a suspicion, and the cost of carrying one around a building where seventeen people work is worse than the money.

What I’d do if I were starting again

Keep every stocktake in one place so you can read category direction across visits rather than one report at a time. Never take a net figure at face value — check the gross movement under it, because mine hid twenty-seven times its own size. Do your line checks and keep them, because that’s the only number you own between visits. And go and look at your till layout with fresh eyes, because that’s where an unreasonable amount of it lives.

Draught Lager +£3,697 across five audits is not somebody carrying kegs out of the back door. It’s a button.

Common questions

What causes stock variance in a pub?

Venners’ observation audits reported in the Morning Advertiser in 2019 attributed 46% of loss to human error and 39% to theft, with losses found at 100% of sites audited. In my own experience the biggest single cause is drinks poured as one product and rung as another, usually because similar products sit next to each other on the till.

How do I tell theft from till errors in a stocktake?

Look at direction across multiple stocktakes rather than the size of one figure. Random variance flips sign; a category negative five audits running is systematic. Errors also tend to pair up — one category consistently down alongside another consistently up — because the stock and the sale went to different places.

Why is my net stocktake result small but I still have a problem?

Because opposite errors cancel out. Across five audits mine showed £11,598 of gross category movement netting down to −£434. On 21 July the net was £12.84 while £1,165 of stock had moved in opposite directions underneath it. Always read the category breakdown, not just the headline.

What yield should I expect on packaged beer?

Venners’ 2021 Yield Report, covering around 22,000 stocktakes across more than 3,000 sites, found packaged beers and ciders averaging 97.6% against an industry expectation of 99.5%, with wine performing best at 98.84%. Venners put up to £10,000 per site per year as potentially recoverable.