Goodhart's Law
When a measure becomes a target, it stops being a good measure.
- When a measure becomes a target, it stops being a good measure.
- Operating: Most call centres are run on 'average handle time.' Agents optimise for short calls by hanging up early, transferring aggressively, or not listening to the customer.
- Investing: Funds judged on quarterly performance trade more aggressively at quarter-end to game the snapshot.
- Everyday life: When you start tracking your steps, you walk more.
British economist Charles Goodhart wrote this in 1975, about monetary policy. The principle has since become one of the most useful general-purpose laws in business.
The pattern: you find a metric that correlates with something you actually care about. You start measuring it. Then you make people responsible for moving it. People immediately start gaming the metric, often in ways that destroy the original correlation. The metric stops measuring what you cared about.
Common operating failures from Goodhart's Law:
- Engineering judged on "lines of code" → engineers write more verbose code
- Customer service judged on "tickets closed per hour" → agents close tickets prematurely, customers come back angrier
- Sales reps paid on "deals closed" → they close any deal, including bad ones, then someone else deals with the cleanup
- Schools judged on "graduation rates" → standards drop until everyone graduates
- A bank judged branches on "new accounts opened" → Wells Fargo employees opened 3.5M+ fake accounts to hit the number
The deeper issue: a metric is always a proxy for what you actually want. The moment people are accountable for the proxy, they optimise the proxy, not the underlying thing. The numbers get better and the underlying thing gets worse.
There's no clean defence, but a few moves help:
1. Use multiple metrics that are hard to game in the same direction. Closing speed AND customer satisfaction. Revenue AND gross margin. 2. Refresh metrics periodically so people can't optimise too cleverly against any one. 3. Audit the metric against the underlying thing once in a while. If the underlying thing is getting worse while the metric is getting better, Goodhart has struck. 4. Be honest with yourself that any single metric will eventually be gamed. Treat metrics as approximate, not as truth.
A related warning. The more important the metric, the harder the gaming. Your most-tracked KPI is probably the most distorted one.
Examples in the wild
Most call centres are run on 'average handle time.' Agents optimise for short calls by hanging up early, transferring aggressively, or not listening to the customer. The metric goes down. Customer satisfaction also goes down.
Funds judged on quarterly performance trade more aggressively at quarter-end to game the snapshot. The metric (Q-end value) becomes the target. Long-term returns get worse. Investors get hurt.
When you start tracking your steps, you walk more. Then you start pacing your apartment at 11pm to hit 10,000 because the number matters more than the walking. The metric is no longer measuring health.
Goodhart's Law is one of the mental models we apply through real cases inside the Pareto MBA — a part-time program for professionals who want to think clearly about business.