The Shift That Always Runs Over

The rota says the shift ends at 03:00. Payroll says 04:20. Every week.

Nobody is slacking. The work genuinely takes that long. The schedule is simply describing a night that does not happen.

Why it never gets corrected

Because the overrun is invisible in the place where schedules are made.

The rota is built from planned hours. The overrun appears in payroll, weeks later, aggregated with everything else. Two different documents, two different people, and neither one shows both numbers side by side.

So the schedule keeps saying 03:00, and the cost keeps arriving at 04:20.

Where the time actually goes

Measured, in most rooms, it splits into four:

The last guests. Not the largest, but the most variable. A table that stays forty minutes past last orders moves everything behind it.

Cash-up. Fixed and predictable, and usually underestimated when the rota was written — frequently by someone who has never done it on a Saturday.

Clean-down. Scales with how busy the night was. A quiet Tuesday and a full Saturday are not the same job, and the rota often gives them the same time.

Waiting. The most fixable and the least noticed. Staff who have finished their own section waiting for one person to finish theirs, or for a manager to check something, or for a taxi.

⚠️ The fourth is usually the largest single block and the cheapest to remove. It is also the one nobody mentions, because from the inside it does not feel like work.

The measurement

Two weeks. One line per shift:

scheduled end
actual end
what the last person was waiting for

The third column is the one that matters and the one people skip. Without it you learn that you run over — which everybody already knew.

What tends to be fixable

Staggered finishes. Not everyone needs to leave at the same time. If three of eight can go at 03:00 and the rest at 03:45, that is a real saving and it needs no process change beyond deciding it.

Moving what can be done earlier. Some of the clean-down can happen during the last hour. Some cannot. Knowing which is which is worth an afternoon of asking.

Removing the checks that create a queue. If one manager must sign off five things, four people wait for one person. Either delegate the checks or sequence them.

Setting last orders so the ending is predictable, and enforcing it consistently. An inconsistent last orders is a schedule that cannot be written.

What is not fixable, and should be scheduled instead

Some of it is simply the work.

If cash-up takes forty minutes, the rota should say forty minutes. Scheduling thirty and paying forty is not a saving — it is the same cost with a worse forecast, and it makes every other number you build on top of the rota wrong.

⚠️ The point of measuring the overrun is not always to eliminate it. Frequently the right outcome is a schedule that matches reality, which costs the same and lets you plan.

The part that costs more than the money

An unpredictable finish is one of the most common reasons people leave hospitality jobs. "I never know when I'm getting home" outranks pay in a lot of exit conversations.

So the overrun is not only a labour cost. It is a retention cost, and that one does not appear on any line item at all.

Three to hold

Median overrun by night of week. Tells you whether it is one night or the whole week.

Overrun on the busiest night versus the quietest. If they are similar, the cause is fixed work scheduled wrong, not volume.

Share of overrun spent waiting. The fixable part.

Where the record has to sit

Planned hours in the rota, actual hours in payroll, and the reason in nobody's system. That is why this runs for years in venues that watch costs carefully everywhere else.

tasteck keeps scheduled and actual shift times on the same record, so the gap is a column rather than a reconciliation someone does at month end.

The schedule is not wrong once a week. It has been wrong every week, and it is the only document nobody checks against the night.