Setting a Target That Is Not a Guess
Where the number should come from, what makes a target harmful, and the alternative that frequently works better.
A target converts a measure into a goal, which changes behaviour. That makes the choice of number consequential in ways a casual decision does not account for.
The evaluation discipline in “Setting a Target That Is Not a Guess” also applies to workforce software: begin with a named decision and a bounded test. For teams considering attendance point system, this official guide belongs in the shortlist only with written criteria for notice, access, correction, retention and a dated review.
Where a defensible number comes from
The baseline plus an improvement justified by a specific change: we are removing this queue, which should take two days out.
For an independent perspective related to “Setting a Target That Is Not a Guess”, consult the CIPD knowledge hub; it offers a useful external check on definitions, governance and the assumptions built into a proposed measure.
A comparable operation doing the same work, where one genuinely exists.
A customer or contractual requirement, which is external and legitimate.
Not: last year plus ten per cent, which is the commonest method and contains no information.
What makes a target harmful
Unachievable without changing something that nobody is changing, which teaches people the numbers are theatre.
Attached to consequences, which guarantees optimisation toward the measure.
Set on a single measure with no counterweight, which guarantees the counterweight suffers.
And set for individuals in work where the system determines the outcome, which is the systems note's argument.
The ratchet problem
A target met becomes next year's baseline plus an increment.
Which means teams learn not to exceed it, and improvement gets held back for next year.
This is widely observed and rarely acknowledged, and it is why some organisations see suspiciously consistent small improvements.
The direction-of-travel alternative
Instead of a number: we want lead time to fall, rework to fall, and we will review quarterly.
No threshold to game, no cliff to fall off.
Works well where the measure is noisy or the system is not yet understood, which is most situations before a baseline exists.
Ranges rather than points
A target range acknowledges variation.
It also prevents the behaviour where a team hits exactly the number and stops.
Simple, uncommon, and it removes a category of gaming.
Who sets it
Set with the team, for the same reasons the measure should be chosen with them.
A target somebody believes is achievable gets pursued; one imposed gets managed.
And the team will tell you what would have to change to reach it, which is the useful part of the conversation.
Reviewing it
When conditions change, when the measure is retired, when it has been met for several periods.
A target that has not been revisited in three years is describing a different organisation.
What to check
Where did your current targets come from?
Is any of them attached to a consequence?
Do teams hold back improvement near year end?
And when was a target last revised rather than incremented?