Labour cost percentage is the most quoted number in shift operations and one of the easiest to misread. The arithmetic is simple: labour cost divided by revenue for the same period, expressed as a percentage. Almost everything difficult about it hides in what you put in the numerator, what you put in the denominator, and what you then compare the result against.
What goes in the numerator
Wages, obviously. After that it gets contested. A figure built from base wages alone isn't comparable with one that includes payroll taxes, employer contributions, holiday accrual and premiums, and the difference between the two versions is not small.
Salaried staff are the other quiet decision. Including a site manager's cost makes the percentage a truer picture of what the site consumes. Excluding it makes the number more responsive to the scheduling choices you can actually change from one week to the next. Both are defensible positions. Only one of them can be in use at a time.
There's no single correct answer here. There's only a correct practice, which is to define it once, write the definition down next to the number, and never quietly change it. A percentage that improves because somebody moved a cost out of the calculation is worse than no percentage at all, because it will be believed.
The denominator moves too
Revenue for the same period sounds unambiguous until you try it. Gross or net of tax. Before or after discounts, comps and refunds. And which period exactly, when the pay period and the trading period don't share a boundary, which they usually don't.
That last one produces the strangest results. If a fortnightly payroll cuts through the middle of a trading week, some of your labour lands in a period whose revenue it didn't generate, and the percentage swings for reasons that have nothing to do with how anybody was scheduled.
It's a ratio, so either side can move it
This is the trap that catches people fastest. The number can improve because labour fell, or because revenue rose, and those are completely different situations.
- A quiet week with unchanged staffing looks like a cost problem
- A busy week with thin staffing looks efficient, and may be service you lost
- A price rise improves the ratio without changing anything operationally
- Cutting hours below what the shift needs improves it, right up until it doesn't
So the percentage is a question rather than an answer. When it moves, the first thing to establish is which side of the ratio moved.
Compare it against itself
Benchmarks drawn from other businesses are mostly noise, because the definitions differ and nobody publishes theirs. Even inside one company, two sites with different opening hours, service models or wage floors aren't directly comparable, and ranking them by this number quietly rewards the site with the easiest conditions.
The comparison worth making is against the same site's own history, and against what that site was asked to deliver. A percentage read next to the coverage requirement it was staffed for tells you something. Read on its own, it mostly tells you what kind of week it was.
Use it as a prompt
The number works best as an alarm rather than a target. It flags a period worth looking at. What actually diagnoses that period is the layer underneath: which hours were overstaffed, where unplanned overtime appeared, and which shifts ran longer than they were scheduled to.
“One percentage can't tell you what happened. It can tell you which week to go and look at.”