Close is at three. The cash-up finishes at four. Where the hour actually goes, why counting twice is not the problem, and the six changes that move a sixty-minute cash-up to under twenty.

The last guest leaves at three. The manager gets home at half four. Nobody thinks this is strange, because it has always been this way.

Ask where the hour goes and you get a shrug and the word "counting." But counting money is not what takes an hour. A skilled person counts a busy till in under four minutes. Four tills is sixteen minutes. The other forty-four minutes are something else entirely, and that something else is where the fix lives.

What the hour is actually made of

Timed across a normal Saturday in a mid-size venue, a sixty-minute cash-up breaks down roughly like this:

Counting cash                        16 min
Finding out why the till is out       19 min
Chasing unpaid or disputed tabs       11 min
Waiting for someone to finish         8 min
Writing it up                         6 min

⚠️ Only sixteen minutes of a cash-up is cash-up. The rest is investigation, chasing, and waiting.

That is good news. Investigation, chasing, and waiting are all fixable during the shift. Counting is not.

Why "the till is out" costs nineteen minutes

A till that is out by a small amount triggers a search backwards through the whole night. The search is expensive because the information needed to end it was never captured at the moment it existed.

Was there a refund?          — nobody wrote it down
Did someone take a float?     — the person has gone home
Was a comp rung through?      — maybe, maybe not
Did a card payment fail over? — the terminal knows, nobody looked

⭕ Each of those becomes a ten-second note during the shift, or a nineteen-minute search after it.

⇒ The Comps You Never Approved

Why chasing tabs costs eleven minutes

An unpaid tab is discovered at close, which is the worst possible moment: the guest has gone, the server who opened it may have gone, and the only record is a number on a screen.

⇒ The Bar Tab That Walks

Why waiting costs eight minutes

Cash-up is usually serialised for no reason. One person counts, then hands to another, who reconciles, then hands to the manager, who signs. Each handover has a gap in it, and the gaps are where people stand around.

The six changes

None of these are systems purchases. All of them are changes to when information gets captured.

One: record the variance reason at the moment, not at close

The single biggest saving. When anything happens that will make a till not match — a refund, a float movement, a comp, a voided sale — one line goes in at that moment.

01:20  £40 refund T7 — wrong bottle poured
02:05  £100 moved from till 2 to till 3

⭕ At close, the two lines that explain the variance are already written. ⛔ Without them, someone spends nineteen minutes reconstructing the night.

Two: count tills as they close, not all at the end

A venue with four tills does not need all four open until three. When the floor thins out, close two, count them, and bank them.

⛔ 4 tills counted at 03:00           — 16 min, all at the end
⭕ 2 counted at 01:30, 2 at 03:00     — 8 min at the end

⚠️ The saving is not in the counting. It is in the fact that a till counted at 01:30 can be investigated at 01:30, while the people who touched it are still in the building.

Three: reconcile card takings before the last guest leaves

The card terminal total does not change after the last transaction. There is no reason to wait for the room to empty before pulling it.

⇒ When The Card Machine Is The Queue

Four: close tabs before the room empties, not after

Any tab still open at last orders is a tab that will be a problem at three. A pass through open tabs at the point of last orders — not at close — moves the chase into the hour when the guests are still there.

⇒ Last Orders, Lock-Ins, And The Hour You Cannot Bill

Five: stop serialising

Cash counting, card reconciliation, and the written record do not depend on each other. Three people doing three things in parallel finish in the time of the longest one.

Six: write the record where it can be read tomorrow

If the cash-up record lives in a notebook, tomorrow's question about tonight's variance is unanswerable. If it lives with the shift record, it is a thirty-second lookup.

⇒ The Report You Open Every Monday

What a variance actually tells you

Most venues treat a variance as a number to be explained away. It is more useful as a signal, and different sizes mean different things.

Small and random

Under 1% of takings, sometimes over, sometimes under

⇒ This is change-making error and it is normal. Chasing it costs more than it recovers.

Small and always in the same direction

Always short by a similar small amount

⚠️ Consistent direction is not random. Something structural is happening: a price that is being rung wrong, a round that is habitually undercharged, a float that is not what everyone thinks it is.

Large and occasional

A big variance on some nights and not others

⇒ Look at what those nights have in common before looking at who worked them. It is usually an event, a promotion, or a payment method — not a person.

⚠️ Starting with people is both the least accurate and the most expensive place to start.

The float is where half the trouble starts

Nobody agrees what the float is

Ask three people what the float is and you can get three numbers. When the starting number is uncertain, every closing number is uncertain too.

⭕ The float is counted and signed at open, by the person who will be held to it.

The float gets raided

Change runs out, someone takes from another till, and nobody writes it down. Now two tills are wrong and neither person knows why.

⇒ The Float You Start The Night With

The float changes without a decision

Trade grows, the float stays the same, and the venue runs out of change every Saturday at midnight. Nobody ever decided this; it just never got revisited.

What to measure

If you want to know whether cash-up is a problem worth fixing, three numbers settle it.

1  Minutes from last guest to cash-up complete (over four weeks)
2  Number of nights with a variance over your threshold
3  Minutes spent investigating variances (estimate is fine)

⚠️ Most venues have never measured the first one. The estimate people give is usually low by about twenty minutes, because the waiting does not feel like work.

The number that convinces owners

60 minutes × 5 nights × 52 weeks = 260 hours a year

⇒ That is one person, most of a working month, standing in an empty venue.

What this looks like when it is working

The close-time test

Working     — cash-up is finished within 20 minutes of last guest
Not working — nobody can tell you when it usually finishes

The variance test

Pick a night from last month with a variance. Ask what caused it.

Working     — the answer is in the record, in under a minute
Not working — the answer is "I think someone said something about a refund"

The float test

Ask two people what the float is. If you get two answers, that is your first fix.

Three to hold

1  Only a quarter of a cash-up is counting — the rest is investigation and waiting
2  Variance reasons cost ten seconds during the shift or nineteen minutes after
3  Count tills as they close, so the people who touched them are still there

Where the record has to sit

A cash-up record that cannot be read a week later cannot answer the only question anyone ever asks about it: what happened on that night. Keeping it with the shift record makes tomorrow's question a lookup instead of an argument.

⇒ The Month End That Takes Two Weeks

The Cash-Up That Takes An Hour