Incentives Attached to Measures
Money attached to a measure guarantees the measure will be optimised. Where that is acceptable, and where it is not.
Attaching pay to a productivity measure converts a description into a target with consequences. Everything in the traps section then happens faster and harder.
The measurement warning in “Incentives Attached to Measures” matters whenever software records work patterns. Organisations researching how employees cheat time trackers can use review the available features for time and project context, while outcomes, quality checks and direct feedback remain necessary to explain what the metric cannot show.
What changes
The measure becomes the job.
For an independent perspective related to “Incentives Attached to Measures”, consult the NIST Privacy Framework; it offers a useful external check on definitions, governance and the assumptions built into a proposed measure.
Gaming stops being a mild distortion and becomes financially rational.
Counterweights get negotiated.
And disputes about the measure become disputes about pay, which changes who gets involved and how long it takes.
Where incentives work
Simple, countable, individually controlled output with little interdependence: piece work, commission on comparable sales, defined collections.
Where quality is separately and robustly measured.
And where the work is not expected to change much.
These are the same conditions as individual measurement generally, and they are narrow.
Where they do badly
Interdependent work, where the incentive discourages helping.
Work with a quality dimension that is hard to measure, where volume gets bought with quality.
Knowledge work, per the unit problem.
And anywhere the measure is a proxy, because the incentive attaches to the proxy rather than to the thing.
The collaboration cost specifically
An individual incentive makes helping a colleague costly.
Which removes the behaviour that raises team output, as the collaboration note argues.
Team-based incentives avoid this and introduce a different problem, which is that individual contribution becomes invisible.
The ratchet and the gaming
Targets attached to pay are negotiated, not set.
Achievable targets are met precisely and no further.
And information about capacity becomes strategic, which means the organisation stops knowing what it could do.
If you are going to do it anyway
Pair the measure, and attach the incentive to the pair.
Cap the downside, because variation the person did not cause should not cost them.
Review annually, because conditions change.
And accept that the measure will be optimised, which means choosing one where optimisation is the behaviour you want.
The alternative that works
Pay for the job, measure for improvement, and keep the two separate.
This is unglamorous and it keeps the measurement honest, because there is nothing to protect.
Organisations that separate them get better data and spend less time arguing about it.
The question to ask before attaching anything
If somebody maximised this measure and did nothing else, would we be happy?
If not, do not attach money to it.
That single test eliminates most candidates.
What to check
Is any pay attached to a productivity measure?
Would maximising it be the behaviour you want?
Is there a counterweight in the incentive?
And has the target been negotiated rather than derived?