Why Sustainable Performance Requires More Than Good Metrics

Where it converges and where it diverges

Read together, these four bodies of research converge on a shared conclusion from four different directions: a metric’s design determines what gets measured, but it does not, on its own, determine what gets produced. Ridgway’s research shows that a metric alone can create unintended distortion. Kerr’s research shows that a metric’s effect depends on the full reward structure around it, not the metric in isolation. Repenning and Sterman’s research shows that a metric’s improvement can be driven by attention rather than durable capability, making the metric’s short-term trajectory an unreliable signal of long-term sustainability on its own. Edmondson’s research shows that a metric’s accuracy over time depends on whether people feel safe contributing honest information to it. None of the four researchers argues against measurement; each identifies a distinct condition, external to the metric’s own design, that determines whether the metric can do the job it was built to do.

Where the research diverges is in what it implies should receive primary attention. Ridgway’s and Kerr’s work point toward incentive and reward design as the central lever — get the incentives right, and the metric will track something closer to what the organisation actually wants. Repenning and Sterman’s work points toward embedded capability as the central lever — build the underlying capacity for the work to sustain its own quality, rather than relying on periodic bursts of attention, regardless of how the metric is defined. Edmondson’s work points toward relational and psychological conditions as the central lever — without safety to surface honest signal, even a well-incentivised, capability-embedded system will eventually lose the accurate information a metric depends on to mean anything at all. These are not contradictory positions so much as different entry points into the same underlying problem, and an organisation that addresses only one — redesigning incentives without building capability, or building capability without addressing psychological safety — is, according to this body of research collectively, likely to find its gains less durable than expected.

A further point of convergence worth naming is what none of these four studies argues: none suggests metrics should be abandoned, simplified into vagueness, or treated with blanket distrust. Each treats measurement as a necessary but insufficient component of a larger system — a conclusion that sits between the two more extreme positions an organisation might otherwise default to, either placing excessive faith in a well-designed metric or dismissing measurement altogether once one has been found to mislead.

The four conditions identified across this research also differ in how visible their absence is likely to be. A measurement-integrity failure of the kind Ridgway documented tends to surface eventually, because the gap between the metric and reality widens until it produces an undeniable discrepancy. An incentive-misalignment failure of the kind Kerr documented can persist for considerably longer, because the behaviour it produces is often fully compliant with every written rule and therefore generates no obvious trigger for anyone to investigate. A capability gap of the kind Repenning and Sterman documented is arguably the hardest to see in real time, since it only becomes visible once the attention sustaining a gain is withdrawn — which may be months or years after the gain was first celebrated. A psychological-safety gap of the kind Edmondson documented may be the least visible of all from a leadership vantage point specifically, since its primary symptom is an absence — information that was never raised — rather than a presence that can be observed and measured directly.

This asymmetry in visibility has a practical implication worth stating plainly: an organisation that only investigates the condition most likely to produce a visible symptom will systematically under-invest in the conditions whose failure stays hidden longest, not because those conditions matter less, but because they are structurally harder to notice until the cost has already accumulated.

Taken together, the four studies suggest that “sustainable performance” is better understood as a property of a system with several interdependent parts than as an attribute of any single well-chosen number. A metric can be well designed, an incentive structure can be broadly sound, embedded capability can be genuinely present, and psychological safety can be reasonably strong — and an organisation that has all four in place is still describing something closer to a set of conditions it has to keep maintaining than a problem it has permanently solved. None of the four researchers reviewed here frames their finding as a one-time fix, and the shared implication of reading them together is that sustainability itself is an ongoing property, not a state an organisation arrives at and then stops tending to.