Charles Goodhart was a British economist. In 1975, he observed something about monetary policy that turned out to apply to almost everything in management.
Goodhart’s Law: “When a measure becomes a target, it ceases to be a good measure.”
The moment you tell a team to optimise for a number, they optimise for the number. Not for what the number was measuring.
The Mechanism
Goodhart’s Law doesn’t require bad intent. Teams don’t game metrics maliciously. They do what rational people do: they optimise for what they’re measured on.
The problem is that metrics are proxies. Engagement rate is a proxy for value. Revenue is a proxy for customer satisfaction. Velocity is a proxy for productivity. And every proxy has ways to improve it that don’t improve what it was measuring.
When the proxy becomes the target, the gap between the number and the underlying reality quietly widens — until one day the number looks fine and the business doesn’t.
What It Looks Like in Practice
You see it in engagement metrics that climb while retention stays flat. Active usage rises because awareness grew, not because the product got stickier. The number moved. The business didn’t.
You see it in velocity metrics that reward closing tickets over solving problems. Teams learn to break work into smaller pieces, merge faster, and close more issues. The throughput numbers are excellent. The customer outcomes aren’t.
You see it in NPS campaigns that exist to generate a score rather than surface honest feedback. The score improves. The insight dries up.
In each case, the team is doing exactly what was asked. Goodhart says that is precisely the problem.
What to Do About It
This is not an argument against metrics. Measurement is how you know anything. But Goodhart points to three disciplines that most orgs skip:
-
Measure at multiple levels simultaneously and hold the tension between them. Measure engagement and renewal. Measure velocity and customer outcomes. Measure NPS and the verbatim feedback that NPS can’t capture. The moment you drop one half of the pair, the remaining measure becomes gameable — and will be gamed.
-
Ask the gaming question regularly. What would a team hitting this number without actually improving the underlying thing look like? If you can describe it clearly, you need a second measure. Build this into your review cadence, not just your retrospectives.
-
Treat the dashboard as a hypothesis, not a verdict. The most dangerous metric is the one everyone trusts without questioning what it doesn’t capture. Every number is an argument. Know what argument it is making and what it cannot see.
The dashboard that makes you feel informed might be the thing that is keeping you uninformed.
Pick your most-watched dashboard metric. What would it look like if your team were hitting it without actually getting better?
The Laws They Don’t Teach series:
Leave a Reply