When the Number Becomes the Goal: Goodhart's Law and the Limits of Measuring People

When the Number Becomes the Goal: Goodhart's Law and the Limits of Measuring People

The moment a measure becomes a target, people optimize the measure instead of the thing it was standing in for. For anything as human as leadership or growth, that's not a risk to manage — it's a reason to be careful what you hand to a number.

Metrics are useful. They track progress, expose problems, and turn vague intentions into something you can actually see move. None of what follows is an argument against measuring things.

It's an argument about what happens when a measure stops being a description and becomes a goal — because at that moment, reliably and across every domain anyone has studied, it starts to corrupt.

The law, and why it's iron

The economist Charles Goodhart noticed the pattern in 1975, watching monetary policy: the Bank of England had found stable relationships between certain money-supply measures and inflation, but the moment it started targeting those measures to control inflation, the relationships broke down. His formulation was dry — "any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes" (Goodhart, 1975) — but the anthropologist Marilyn Strathern later gave it the phrasing everyone quotes: when a measure becomes a target, it ceases to be a good measure (Strathern, 1997).

The mechanism is simple and it doesn't require anyone to be dishonest. A metric is a proxy — a measurable stand-in for something you actually care about but can't observe directly. Test scores stand in for learning. Call-handling time stands in for service. Engagement scores stand in for a healthy culture. As long as no one's optimizing the proxy, it tracks the real thing reasonably well. The instant you attach stakes to the proxy, people rationally pour their effort into moving it — and the proxy and the real thing come apart. Schools teach to the test. Call centers hang up faster. The number improves while the thing it was supposed to represent gets worse.

This was independently discovered from another direction. The social scientist Donald Campbell formulated what's now called Campbell's Law: the more any quantitative indicator is used for high-stakes decisions, the more subject it becomes to corruption pressure, and the more it distorts the process it was meant to monitor (Campbell, 1979). Two researchers, different fields, same iron conclusion — which is part of why it's worth taking seriously. This isn't a management fad. It's closer to a law of the systems.

The part that matters most for leadership

Here's where it stops being a general business insight and becomes specific to the work of developing people, because this is where the corruption is hardest to see and most costly.

The most valuable things a leader does are the least quantifiable. Judgment. The quality of a decision made under uncertainty. Whether people feel safe enough to bring bad news. Whether the person is actually growing or just performing growth. These are real — they show up in outcomes eventually — but they resist being captured in a single number, and the moment you try to force them into one, Goodhart's Law does its work.

Measure a leader on engagement scores, and you may get managers pressuring their teams to rate highly rather than building anything worth rating. Measure coaching by self-reported confidence, and you can manufacture confidence without touching the judgment underneath it — sometimes the confidence rises precisely because the hard self-examination didn't happen. Measure therapeutic progress by symptom-relief checklists tied to reimbursement, and you optimize for the checklist, not the deeper change. In each case the number goes the right direction while the real thing stalls or reverses, and the number is now actively hiding that fact.

This is the reductionism trap, and it's not solved by finding a better single metric. Leadership, resilience, and psychological growth are genuinely multi-dimensional — a person and the system they operate in, interacting — and any attempt to collapse that into one tracked figure doesn't just lose information. It changes behavior toward whatever the figure rewards, which is rarely the thing you actually wanted. A confidence score measures learned responses as readily as real self-awareness, and once it's the target, it will increasingly measure the former.

What to do instead

The answer isn't to abandon measurement and run on vibes. It's to keep metrics in the role they're actually good at, and refuse to promote them past it.

Keep the measure a measure, not a target. A metric works best as an indicator you watch — a signal that prompts a question — rather than a goal you optimize. The moment a number becomes the thing people are rewarded for moving, assume it's begun to decouple from what it was tracking, and watch for the gap.

Use several, deliberately in tension. A single metric is easy to game; a set of measures that pull against each other is much harder. Speed against quality, short-term against long-term, the quantitative number against a qualitative read. The tension is the point — it makes gaming one measure show up as damage in another.

Protect the qualitative. The most important information about a leader or a piece of work is often the unquantifiable kind — is the judgment sound, is the person actually more capable, would you trust them with the harder version of this. Treating those as "soft" and deferring to whatever's easy to count is exactly how Goodhart's Law wins. The numbers inform the judgment; they don't replace the person making it.

Ask what you're actually optimizing for. Behind every metric is a real outcome it was meant to approximate. When the metric and the outcome diverge — the score is up but the thing isn't better — the outcome is what matters, and the metric is what's lying.

The reframe

Goodhart's Law isn't a caution against numbers. It's a caution against letting a number quietly become the goal, at which point it stops measuring what you cared about and starts measuring how well people can move it. That's true everywhere, and it's most true for the things that matter most and quantify least — judgment, growth, the health of a culture, the actual development of a leader.

The useful posture is to hold measurement lightly: to use metrics as a compass that tells you roughly where you are, and never to mistake them for the destination. Real success isn't what the number says. It's what actually changed underneath it — which is often precisely the thing no single number was ever going to capture.

References

Campbell, D. T. (1979). Assessing the impact of planned social change. Evaluation and Program Planning, 2(1), 67–90.

Goodhart, C. A. E. (1975). Problems of monetary management: The U.K. experience. In Papers in Monetary Economics(Vol. 1). Reserve Bank of Australia.

Strathern, M. (1997). "Improving ratings": Audit in the British university system. European Review, 5(3), 305–321.