Campbell's Law
The more a metric drives important decisions, the harder people will work to distort it.
Why PMs should care
Goodhart's Law is about whether the measurement itself still means anything. Campbell's Law is about what happens when that measurement starts deciding people's careers.
Once a metric controls promotions, headcount, or budget, expect it to be gamed on purpose — not quietly, but strategically, with slides and talking points. And the distortion builds on itself: the team that games the metric best gets rewarded, the team that reports honestly ends up looking worse by comparison, and within two quarters the honest team has learned to game it too. Not because they've given up — because they've seen what happens to the people who don't.
The only real defence is to change how you evaluate people often enough that no single way of gaming it stays useful for long, or to pair the gamed metric with a qualitative check — something a human actually reads and makes a judgement on.
Example in product work
A company decides DAU is how product teams will be evaluated for headcount in the next planning round. Within one quarter, you see:
- a push notification sent at 7am with the subject 'Your weekly summary' that counts as a session if opened
- a splash screen that logs a 'view' even if the app is opened accidentally
- a widget that auto-refreshes on the home screen and pings a DAU event
- a redefinition of DAU to include anyone who received a notification, not just opened it
The DAU line on the leadership dashboard is up and to the right. Retention is flat. Engagement per session is down. The team that was actually improving the product has been out-manoeuvred by the team that was optimising the definition.
What to do when you see it
- Goodhart's Law is about the measurement. Campbell's Law is about the careers that depend on it.
- Once a metric decides promotions, expect it to be gamed strategically — with slides and talking points, not accidents.
- Honest teams end up looking worse than gaming teams, which teaches them to game too. It's a predictable spiral.
- Change how you evaluate people often enough that no single gaming strategy stays useful — or pair the metric with a human judgement.
Sources & further reading
- Assessing the Impact of Planned Social Change — Campbell, 1976The original paper where Donald Campbell stated the law about quantitative social indicators.
- Campbell's Law — TechTargetA concise, well-sourced explainer covering the theory, examples, and how it differs from Goodhart's Law.