Utilisation, and Why High Is Not Good
Keeping everybody busy is the instinct and it makes the system slower. The arithmetic behind that is well established.
Utilisation measures how much of available capacity is in use. Managers push it toward full, and the effect on delivery is the opposite of what is intended.
The measurement warning in “Utilisation, and Why High Is Not Good” matters whenever software records work patterns. Organisations researching monitask pricing can use Monitask pricing and plan details for time and project context, while outcomes, quality checks and direct feedback remain necessary to explain what the metric cannot show.
The arithmetic
As utilisation approaches full capacity, queue length rises sharply rather than gradually.
For an independent perspective related to “Utilisation, and Why High Is Not Good”, consult the OECD productivity resources; it offers a useful external check on definitions, governance and the assumptions built into a proposed measure.
At high utilisation, a small increase in demand produces a large increase in waiting.
This is a property of queues rather than of people, and it holds in roads, hospitals, call centres and offices alike.
What it looks like in practice
A team at eighty per cent has slack: unexpected work absorbs.
At ninety-five, everything queues and lead times lengthen dramatically.
And at full, the system stops responding to anything new, which is experienced as the team being unresponsive rather than as being overloaded.
The manager's instinct
Idle capacity looks like waste.
Which is true in a single-step process with predictable demand and false everywhere else.
Variability is what makes slack necessary, and all real work has variability.
The specialist case
The person only one of whom exists: the approver, the expert, the person who knows the old system.
Their utilisation is the one that matters, and it is usually highest.
Which makes them the constraint, and the constraint note covers what to do about it.
Measuring it honestly
Utilisation per person is nearly meaningless in knowledge work, because the denominator is unknowable.
Utilisation of a shared resource — a machine, a queue, a specialist step — is measurable and useful.
Measure the resource rather than the person, which also avoids every fairness problem in the people section.
What to target instead
Lead time, which is what the customer experiences.
Where lead time is acceptable, utilisation is irrelevant.
Where it is not, the answer may be more capacity, less variability or fewer queues — and raising utilisation is none of those.
The counter-case
In capital-intensive work, expensive equipment genuinely should run.
Here utilisation is a legitimate primary measure, and the queueing cost is a known trade accepted deliberately.
The error is importing that logic into work where the expensive resource is people's attention.
What to check
Is anybody targeting utilisation in your organisation?
What is the utilisation of your busiest specialist step?
Does lead time rise when the team gets busier?
And is there any slack in the system at all?