In 1975, the British economist Charles Goodhart, then an adviser to the Bank of England, formulated an observation about monetary policy that would end up transcending his field. Years later, the anthropologist Marilyn Strathern condensed it into the phrase quoted everywhere today: "when a measure becomes a target, it ceases to be a good measure".
The idea is counterintuitive but solid. A metric works as a signal because it captures, imperfectly, something that truly matters but that you can't measure directly. Visits measure interest; followers, influence; sign-ups, demand. As long as you only observe them, they're useful. But the moment you turn them into the target to maximize, people —you included— start optimizing the number itself, and the number separates from what it stood for.
The classic example is the Soviet factory asked to produce nails: measured in number of nails, it made millions of tiny useless nails; measured by weight, it made a single giant nail. The indicator was met perfectly while the purpose —useful nails— was completely betrayed. For whoever builds a business guided by metrics, the trap is the same and quieter.
The naive reading is that measuring is always good: "what gets measured gets improved, so I set a metric, chase it with everything and go for it". And you launch into maximizing sign-ups, followers or downloads as if they were the end.
That's exactly where it goes wrong. Because the moment that metric becomes the goal, you start doing whatever it takes to raise it —even if that empties of meaning what the metric was supposed to represent. You raise the number and lose the thing.
This isn't about stopping measuring.
It's about how confusing the indicator with the goal makes you optimize the shadow and lose the body that cast it.
1. The metric is a proxy, not the thing. You can never directly measure "having a good business"; you measure approximations: revenue, active users, retention. Each captures a part and leaves out the rest. As long as you remember that, they serve you. The error is forgetting they're approximations and treating them as the real goal: then you optimize the part the metric sees and neglect the part it doesn't, which is usually exactly where the value was.
2. The moment you chase it, the cheap way to raise it appears. Every metric has a shortcut that inflates it without creating what it stood for. Want more sign-ups? Give something away and thousands sign up who'll never return. More followers? Buy them or bait them with noise. The number rises, you think you're advancing, and the real business —people who pay and return— doesn't move. You've learned to make tiny nails.
3. That's why you have to change the metric when it becomes a goal, or anchor it to the behavior that truly matters. The defense isn't to stop measuring, but to choose metrics hard to fake —the ones that only rise if the real value rose, like recurring revenue or long-term retention— and rotate or cross-check the rest before the shortcut corrupts them. A metric you can't inflate without creating real value is the only one that survives becoming a goal.
Think of the metric you chase most right now in your project. Sign-ups, followers, visits, downloads, whatever it is.
If you had to raise it 50% this week at any price, how would you do it? And that how, would it create real value or just inflate the number?
If a shortcut to raise it without the business actually improving comes to mind, that metric no longer serves you as a goal: it's measuring your skill at inflating it, not the health of what you think it measures.
If turning a measure into a goal corrupts it, then choosing well what you chase is an advantage over whoever runs after the easy number. The layers:
Between two indicators, choose the one you can't inflate without creating real value. Recurring revenue and 90-day retention are hard to fake; sign-ups and followers, trivial. What truly matters is usually the most expensive to simulate.
You can watch many indicators to understand, but reward (in your focus, your team, your incentives) only those that don't corrupt when chased. Observing is safe; turning into a goal isn't.
Every time a metric rises, ask whether the thing it stood for rose. More sign-ups but the same sales: the number fooled you. Crossing the proxy with the real result unmasks the tiny nail before you fill the factory with them.
Before setting a numeric goal, ask "how could this rise without the business improving?". If the answer is easy, that metric can't withstand being a goal: change it or anchor it to real behavior. Whoever chooses what to measure with this question avoids years chasing numbers that rise while the business doesn't.
The metric was useful when you only watched it. It corrupted when you turned it into the goal. Because the moment something becomes the number to raise, the way to raise it appears that betrays exactly what the number was meant to protect. Measure to understand, not to chase. And if you're going to chase a number, let it be one you can't inflate without, along the way, building the real value it claims to measure.