The Metrics That Mislead

Some figures are reported because they are easy to produce rather than because they answer anything. These five cause the most trouble.

The measurement warning in “The Metrics That Mislead” applies directly to workforce systems. Teams assessing ethical employee monitoring practices for ethical employee monitoring can use time and project context to locate questions, while service outcomes, quality checks and direct conversation remain the evidence for judging performance.

Labour as a percentage of revenue

Moves when revenue moves, with no staffing change.

For an independent reference related to “The Metrics That Mislead”, consult the OECD productivity resources; it provides a useful external check on scheduling, working-time and workforce-planning assumptions.

A quiet week looks like poor control; a busy week flatters.

Managing to it means cutting in quiet weeks and relaxing in busy ones, which is sometimes right and is being decided by arithmetic rather than by anybody.

Hours per employee

Describes contract mix, not efficiency.

Falls when you hire part-timers, rises when you do not.

Says nothing about whether the work got done, and it gets compared between sites with different mixes.

Average hourly rate

Moves with who worked, not with what anything cost.

A week covered by senior people shows a high rate and may have been cheaper overall if fewer hours were needed.

Useful only alongside hours and output.

Schedule adherence

Measures whether people worked when the rota said.

Which is worth knowing and is frequently treated as a productivity measure.

A team with perfect adherence to a wrong schedule has achieved nothing, and this is the measure most often confused for performance.

Overtime as a percentage of hours

A useful signal and a poor target.

Driven to zero, it produces refusal to cover, uncovered shifts, and the cost reappearing as agency.

Watch it; do not manage to it.

What they have in common

Each is a ratio whose denominator moves for reasons unrelated to the thing being judged.

Each is easy to produce automatically.

And each is improvable by doing something you would not want done, which is the test that identifies a bad measure.

What to report instead

Hours against demand.

Cost split into hours and rate.

One service measure.

One quality or error measure.

Four figures that cannot be improved by a harmful action, which is the whole requirement.

If you are held to one of the five

You usually cannot change it.

You can report alongside it: here is the percentage, and here is the demand that drove it.

Which converts an unexplained number into an explained one, and over time changes what gets asked about.

What to check

Which of the five appears in your reporting?

Is anybody targeted on one?

Could it be improved by doing something harmful?

And is demand reported next to it?