When a measure becomes a target, it stops being a good measure.
Once people optimize for the metric, the metric stops measuring what you cared about.
Pick any KPI you've ever cared about. The moment it became someone's bonus, they found the cheapest path to the number — not the result the number was supposed to proxy.
The fix isn't a better metric (there isn't one). The fix is to balance measures, rotate them, and never bonus a single number that's easy to game.
The applied version.
Goodhart's Law is the law of unintended optimization. It says that every metric is a proxy — a thin slice of the outcome you actually want — and that any proxy bonus-linked long enough will detach from the outcome and live its own life.
The deeper version, from Marilyn Strathern, is sharper: "When a measure becomes a target, it ceases to be a good measure." That word "ceases" is the warning. It's not that the metric becomes mildly less useful — it stops measuring the thing at all. NPS becomes a survey design exercise. Sales pipeline becomes a CRM-filling exercise. Lines-of-code becomes a comment-padding exercise. The metric reports green while the underlying reality goes red.
The three defenses are well established and almost never used: (1) balance — pair every KPI with a counter-metric that catches the cheapest cheat (speed paired with quality, volume paired with retention); (2) rotate — change which metric is bonus-linked each cycle so no one can build a career out of gaming one; (3) narrate — require people to explain how they hit the number, not just whether they did.
The strategic point: most boards still ask for a single "north-star metric." That request is itself a perverse incentive. Every CEO who has been burned knows this — the question is whether you can build the courage to push back and design the measurement system properly.
Soviet nail factories rewarded by weight produced one giant nail. Rewarded by count, they produced thousands of tiny ones. The metric was perfect. The output was useless.
You leave Week 4 with a Goodhart-proof scorecard for your team — balanced, rotating, narrated — that we'll defend with you in a mock board review.
- →The Balance / Rotate / Narrate framework with worked templates for 6 functions.
- →30+ real-world Goodhart failures dissected (Volkswagen emissions, Wells Fargo, Veterans Affairs, COMPAS).
- →Mock board defense: live-fire critique of your scorecard by Aaron Bare and a fellow cohort member.
Your support team's bonus depends on average call time under 4 minutes. What probably happens?
Pick a metric you (or your org) track. Write the laziest, cheapest, most embarrassing way to hit it without delivering the underlying outcome.
Hint: If you can think of the loophole in under 60 seconds, your team already has.