Measurement Dysfunction

Definition

Measurement dysfunction is the failure mode where measuring and managing to a metric distorts behavior away from the outcome the metric was meant to serve. It is the caution behind the popular maxim “what gets measured gets managed” — a phrase that is both misattributed and, taken as gospel, misleading.

Core Ideas

The maxim is misattributed and known to be flawed

• The line is routinely credited to Peter Drucker, but — per the Drucker Institute itself — he never said it.
• Its flaws were documented long ago: V. F. Ridgway’s 1956 paper, “Dysfunctional Consequences of Performance Measurements.” Simon Caulkin’s summary sharpens it: “what gets measured gets managed — even when it’s pointless to measure and manage it, and even if it harms the purpose of the organisation to do so.”

Two hard truths

Not everything that matters can be measured. Not everything that we can measure matters.

Measuring what you actually care about is often difficult or impossible, so teams substitute an easy proxy — and then optimize the proxy at the expense of the goal.

Target-gaming in practice

The same brittleness AI researchers see in reward-hacking (algorithms satisfying a metric while violating its intent) shows up in organizations:

• An outsourced police control room placed 999 calls during quiet periods to hit a target of answering 92% of calls within 10 seconds.
• UK doctors were diverted from seriously ill patients to minor cases to meet a 4-hour waiting-time target.

Notably, some organizations achieve outstanding results without measuring-and-managing the outcome of interest at all (e.g. the Dutch home-care org Buurtzorg — higher patient satisfaction at lower cost).

The practical takeaway

Before adopting a metric, ask whether it truly proxies the goal and what behavior it will incentivize. Start from goals, not from what is easy to instrument — the discipline that Developer Productivity Measurement encodes via Goals-Signals-Metrics and by measuring experience rather than performance.

Relationships

• Developer Productivity Measurement — a positive example of measuring carefully (goals-first, experience not performance) to avoid this failure mode
• OKR — goal-setting frameworks are where badly chosen key results turn into target-gaming
• Respect Over Diversity — a demographic count standing in for the culture it was meant to indicate
• Constructive Feedback (BIQ) — the same failure at conversation scale: a vague label passed off as a judgment

References

• What gets measured gets managed - It is wrong and Drucker never said it