# What gets measured usually gets the wrong thing done faster.
Everyone knows the Drucker line: "What gets measured gets managed." It's one of those quotes that gets nodded at in every boardroom and questioned in almost none. Which is a shame, because it's only half true, and the half it's missing is the half that matters.
What actually happens when you measure something is more subtle and more dangerous: what gets measured gets optimized. And the thing people optimize is the metric, not the outcome the metric was supposed to represent. The dashboard becomes the product. The map replaces the territory. And everyone feels productive while the actual goal drifts further away.
A metric is a simplification. The moment you forget that, you start managing the simplification instead of the reality.
The call center problem
This is not a theoretical concern. Let me give you an example I've seen play out three separate times in different industries.
A call center measures average handle time — how long each call takes. The goal is efficiency. Makes sense. Shorter calls mean more customers served, lower costs, better throughput. The metric is clean, measurable, and easy to put on a dashboard.
So agents start optimizing for handle time. They rush conversations. They transfer problems they could solve because transferring is faster than fixing. They develop scripts that move the call toward closure regardless of whether the customer's actual problem is resolved. Handle time drops. The dashboard looks beautiful. Management presents the numbers at the quarterly review.
And then customer satisfaction falls off a cliff. Repeat call volume goes up, because the first call didn't actually solve anything. The cost of a "resolved" call doubles because each issue now takes two or three contacts. Net promoter score tanks. But handle time — the metric everyone is watching — looks great.
The metric didn't measure the wrong thing. It measured one right thing and ignored everything else. The problem is that when you put a number on a wall and tie it to performance reviews, people will hit that number. They're not being dishonest. They're being rational. You told them what mattered, and they believed you.
Goodhart's ghost
There's a formal name for this — Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure." An economist named it, but every operator has lived it.
I worked with a logistics company that measured on-time delivery percentage. Reasonable metric. The operations team started classifying borderline deliveries as "on time" by adjusting the definition of the delivery window. Late by an hour? That's still the same business day. Late by a day? Well, the customer didn't complain, so we'll mark it as acceptable. The on-time percentage climbed to 97%. The actual delivery experience didn't change at all.
Nobody sat in a room and said "let's manipulate the data." They gradually, rationally, incrementally adjusted their interpretation of the metric to make the number more comfortable. That's how Goodhart's Law works in practice — not through fraud, but through drift. Small definitional adjustments that individually seem reasonable and collectively make the metric meaningless.
What to measure instead
I'm not against measurement. I'm against lazy measurement — picking a number because it's easy to count and then treating it like the truth.
Good measurement practices look different:
Measure the outcome, not the activity. Don't measure how many calls were handled. Measure how many problems were resolved on the first contact. Don't measure how many features were shipped. Measure how many customers used them a month later.
Use pairs of metrics that hold each other accountable. Speed and quality. Volume and satisfaction. Cost and retention. A single metric always gets gamed. A pair of opposing metrics creates a tension that's closer to the actual tradeoff you're trying to manage.
Keep the metric but kill the target. Let the team see the number. Let them discuss it. Let them use it as one input among many. But the moment you say "hit 95% or else," you've turned a thermometer into a thermostat, and people will find creative ways to hold it at 95% regardless of the actual temperature.
Rotate what you measure. This sounds counterintuitive, but hear me out. If you measure the same thing for long enough, the organization learns to optimize for it. Rotating attention — this quarter we're focused on throughput, next quarter we're focused on accuracy, the quarter after that we're looking at customer effort — keeps people focused on the whole picture instead of one pixel.
The dashboard trap
The most dangerous moment in any organization's relationship with metrics is when someone builds a dashboard. Dashboards are extraordinarily useful. They're also an implied hierarchy of importance. The numbers on the screen become the organization's definition of reality.
What's not on the dashboard doesn't exist. Customer effort? Not on the dashboard. Employee morale? Not on the dashboard. The quality of handoffs between departments? Not on the dashboard. These things are real and important but they're hard to count, so they get crowded out by things that are easy to count.
I've started asking a specific question in every engagement: "What's the most important thing happening in your operation that doesn't appear on any dashboard?" The answers are always illuminating. They're usually the things that actually determine whether the organization succeeds or fails.
Count what counts
Measurement is a tool, not a truth. It tells you something useful about your system, but it never tells you everything. The moment you mistake the measurement for the reality, you start managing a fiction — efficiently, precisely, and with beautiful charts.
The best operators I know use metrics the way a pilot uses instruments — as inputs that inform judgment, not as substitutes for it. They look out the window too.