Imagine you want to get healthier, so you buy a step counter and aim for 10,000 steps a day. Six months later, you're pacing your living room at midnight to hit the number—but you haven't improved your diet, sleep, or stress levels. You optimized for the metric, not the goal.

This is the proxy metric trap, and it shows up everywhere: in business dashboards, school grading systems, and personal productivity apps. We measure what's easy to count, then slowly forget it was only ever a stand-in for the thing we actually cared about. Let's investigate how this happens and what better measurement looks like.

Real goals are messy. "Customer satisfaction," "employee well-being," "learning"—these matter enormously, but they're hard to pin down with a single number. So we reach for the nearest measurable thing instead. Customer satisfaction becomes Net Promoter Score. Learning becomes test grades. Health becomes BMI.

The problem isn't that proxies are useless. They often start as reasonable approximations. The danger is what happens next: the proxy quietly replaces the goal in everyone's mind. Teams stop asking "Are customers happy?" and start asking "Is NPS up?" The map becomes the territory. Once that substitution happens, you lose the ability to notice when the proxy and the real goal drift apart—which they almost always do.

Think of it like navigating by a landmark. A water tower on the horizon is a fine guide when you're far away. But if you stare only at the tower, you'll walk straight into a ditch at your feet. Convenient metrics are distant landmarks. They point you in roughly the right direction, but they can't tell you about the ground beneath you. The more weight you put on them, the less attention you pay to the reality they were supposed to represent.

Takeaway

Whenever a metric feels easy to track, ask: what important things about my real goal does this number fail to capture? The gap between the proxy and the goal is where problems hide.

British economist Charles Goodhart observed something deceptively simple: when a measure becomes a target, it ceases to be a good measure. The moment people know they're being evaluated on a specific number, behavior shifts to optimize that number—often at the expense of the underlying goal.

Examples are everywhere once you start looking. A call center measures "average handle time" to improve efficiency, so agents rush callers off the phone—and customer problems go unsolved. A school ties funding to standardized test scores, so teachers narrow the curriculum to test prep—and broader learning suffers. A software team tracks "lines of code written," so developers write verbose, bloated code—and the product gets worse. In each case, the metric moved in the "right" direction while the real objective moved in the wrong one.

This isn't because people are dishonest. It's because humans are extraordinary optimizers. Give someone a clear target and they'll find the shortest path to it. If the target is a proxy, the shortest path usually involves gaming the proxy rather than improving the real thing. The better someone is at optimization, the faster the proxy breaks. It's almost paradoxical: the more seriously you take a proxy metric, the less useful it becomes.

Takeaway

Goodhart's Law isn't about bad actors cheating the system. It's about rational people responding to incentives. If your metric can be improved without improving the real goal, someone will eventually find that shortcut.

So if proxies are dangerous and real goals are hard to measure, what do you do? The first step is to use multiple metrics that triangulate on the real goal. No single number captures something complex, but three or four imperfect measures, viewed together, create a much richer picture. If customer NPS is up but complaints are rising and repeat purchases are falling, the dashboard tells a story that no single metric could.

The second step is to build in "counter-metrics"—numbers that would move in a bad direction if your primary metric were being gamed. If you track call handle time, also track callback rates. If you measure code output, also measure bug counts. Counter-metrics act like guard rails that keep optimization pressure from running off a cliff.

Finally, and this is the hardest part: regularly revisit whether your metrics still reflect reality. Schedule it. Put "do our numbers still mean what we think they mean?" on the quarterly agenda. Metrics decay. The relationship between a proxy and a goal can weaken over time as conditions change and people adapt. Treating measurement as an ongoing investigation—not a one-time setup—is what separates analysts who find truth from those who just produce reports.

Takeaway

Good measurement isn't about finding the perfect metric. It's about assembling an honest constellation of imperfect ones—and staying curious about whether they still point where you think.

Every number on a dashboard is a simplification of something more complex. That's not a flaw—it's a necessity. The flaw is forgetting the simplification happened. The best analysts hold their metrics loosely, always asking what's not being captured.

Next time you're tracking a goal, try this: write down what success actually looks like in plain language, then choose your metrics. If the words and the numbers ever start telling different stories, trust the words. The data serves the question, not the other way around.