It arrives as one monthly number and feels like something that happens to you. It is the largest line you actually control, and almost nobody looks at it in the unit where the decisions are made.

Wages arrive once a month as a single figure. It is usually the largest cost the business has and it feels like weather — something that happens, that you notice, and that you have limited say in.

Almost all of it was decided by somebody writing a rota three weeks earlier, in fifteen minutes, without a number in front of them.

⚠️ What this page cannot tell you

⛔ What you must pay, what counts as working time, what breaks and premiums apply, and how anybody should be engaged are employment law questions with specific answers where you are.

⛔ This page is not about any of that. ⭕ Ask your accountant and an employment adviser — and do it before designing a rota around an assumption.

⭕ What follows is the operational half: seeing the cost in the unit where it is actually decided.

Why the monthly figure hides it

Four reasons, and they compound.

The decision unit and the reporting unit are different

⇒ ⭕ Rotas are written per week and per shift; the bill arrives per month. By the time you see the number, twenty separate decisions have been aggregated into one figure with no way back.

It is compared to nothing

⚠️ ⭕ A wage figure on its own is not information. Against takings it means something; against last month with a different number of trading nights it means almost nothing.

The quiet nights are invisible inside it

⭕ A Tuesday overstaffed by two people costs the same as a Saturday understaffed by two saves — and both disappear into the same monthly total.

⇒ The Slow Tuesday You Keep Open

And the rota is written under time pressure

⇒ ⭕ Fifteen minutes, from last week's rota, with the constraint being who is available. Cost is not in the room when the decision is made, which is why it is not in the outcome.

⇒ The Rota Everyone Screenshots

The unit that actually helps

One change: look at it per shift, against that shift's takings.

Cost of the shift, as a share of what the shift took

⭕ This is the whole method. It turns a monthly abstraction into twenty comparable numbers, and the outliers are immediately obvious.

It will be very different by night

⚠️ ⭕ A Tuesday and a Saturday will not be close, and they should not be. The point is not one target — it is knowing which nights are far from where you would want them.

And the first hour and last hour have their own version

⇒ ⭕ Opening two hours early adds a known cost against a knowable take. Without the per-shift view this is a matter of opinion; with it, it is arithmetic.

⇒ The Night You Break Even

What moves it that nobody notices

Five things, all of which are invisible in a monthly figure.

The shift that starts an hour before it needs to

⭕ Multiplied by everybody rostered. An hour of setup that could be forty minutes is one of the largest recoverable amounts in most venues, and it never appears as a decision.

The last hour with everybody still on

⇒ ⭕ Staggered finishes are the counterpart of staggered starts and are much less commonly used, because sending somebody home early feels like a slight.

The person rostered because they were rostered last week

⚠️ ⭕ Rotas are copied. A shift that made sense in December is still there in March, unexamined, because nobody rebuilds a rota from zero.

The cover that was booked and not cancelled

⇒ ⭕ A quiet Saturday with agency cover booked three weeks earlier. The decision point was the day before and nobody had a reason to look.

⇒ The Agency Staff You Booked For Saturday

And the training that happens by accident

⇒ ⭕ Two people doing one job because one is new. Correct and necessary — but it is a cost that should be visible and planned rather than discovered.

⇒ The Shift Where Everyone Is New

Where cutting it goes wrong

Four ways this becomes expensive rather than cheaper.

Cutting the people who make the money

⚠️ ⭕ The bar and the floor at peak are where spend happens. A cut there reduces takings faster than it reduces cost, and the ratio gets worse.

Cutting until service gets slow

⛔ ⚠️ Slow service is a spend problem before it is a complaint problem. People buy less when buying is harder, and the effect appears in average spend rather than in feedback.

⇒ The Average Spend That Is Not An Average

Cutting the same people every time

⇒ ⭕ Whoever is easiest to send home is not the same as whoever is least needed. Doing it repeatedly to the same people is how a rota problem becomes a retention problem.

And cutting hours instead of fixing the shape

⭕ The cost is usually in the shape — when shifts start and end — rather than in the total. Reshaping is invisible to the team; cutting is not.

The half hour that finds most of it

One: take four weeks of shift costs and takings

⇒ ⭕ Both numbers per shift. You almost certainly have both already, just not next to each other.

Two: sort by the ratio

⚠️ ⭕ Two or three shifts will be conspicuously worse than the rest. Those are the whole finding, and they are usually not the ones anybody would have guessed.

Three: look at the start and end times, not the headcount

⭕ Most of the difference is in the shape rather than in the number of people. This is also the change least noticed by the team.

Four: rebuild one rota from zero

⇒ ⭕ Not by editing last week's. Once, from the trade pattern. It takes an hour and it is the only way to see what has accumulated.

Five: put the two numbers side by side permanently

⇒ The Report You Open Every Monday

What to protect while doing it

Three things worth deciding not to cut.

The person who holds the shift together

⭕ ⚠️ Every rota has one, and they are rarely the most expensive. Removing them to save an hour is the most expensive hour you will ever save.

⇒ The Person Everyone Asks First

The overlap at the handover

⇒ ⭕ A shift change with no overlap moves the cost to the following night, in the form of things not passed on.

And the hours people rely on

⛔ ⚠️ Reducing somebody's regular hours has consequences that are not only commercial — how it may be done is a question for your adviser, and how it lands is a retention question either way.

⇒ The Cost Of A Bad Hire In A Small Team

What good looks like

The per-shift test

Ask what last Tuesday cost in wages and what it took. ⇒ Two numbers means this is managed. A monthly figure means it is not.

The shape test

⚠️ Ask when the first person starts and why that time. "Because that is when we start" is the answer this page is about.

The rebuild test

Working     — a rota has been built from zero in the last six months
Not working — every rota is an edit of the previous one

Three to hold

1  The decision is per shift; the report is per month — that gap is the whole problem
2  Most of the recoverable cost is in the shape, not the headcount
3  Cutting at peak reduces takings faster than cost, and the ratio gets worse

Where the record has to sit

Cost and takings for every shift, side by side, with start and finish times — not a monthly total and a separate rota. Kept there, four weeks shows two shifts that are conspicuously wrong and the fix is usually a start time. Kept apart, the largest controllable cost in the business arrives once a month as a figure with nothing to compare it to, decided three weeks earlier by somebody with fifteen minutes and no number in front of them.