The Metric You Chose Is Now Managing You

A support organization was measured on average resolution time and improved it by thirty percent in two quarters. Leadership celebrated. Then customer satisfaction fell, because agents had learned to close tickets quickly and reopen them under new numbers when the customer came back.

Nobody cheated. Everyone did exactly what the measurement asked. This is the normal behavior of measured systems, and it should be designed for rather than discovered.

Assume the metric will be optimized literally

Before adopting a measure, ask the uncomfortable question: what is the laziest way to make this number move? Then assume someone will find it, not out of bad faith but because they are busy and the number is what they are judged on.

• Speed metrics produce shortcuts in quality.
• Volume metrics produce small, easy units of work.
• Satisfaction metrics produce pressure on customers to score generously.
• Utilization metrics produce busy people and unstaffed priorities.

None of these mean the metric is useless. They mean it needs a counterweight.

Pair every metric with its natural opposite

The single most effective practice in metric design is pairing. Resolution time with reopen rate. Sales volume with implementation margin. Delivery speed with defect escape rate. Cost per unit with customer retention.

Pairs are hard to game because the shortcuts that move one degrade the other. Review them together, always, in the same forum. A dashboard that shows only one side of the pair will produce the distortion within two quarters.

Any metric reported alone will eventually be optimized at the expense of something you failed to measure.

Separate the ones you steer by from the ones you judge by

Some numbers are diagnostic: they tell you where to look. Others are evaluative: they determine ratings and pay. Using a diagnostic metric for evaluation destroys its diagnostic value, because people begin managing the number rather than reporting the reality.

Say which is which. Teams will tell you the truth about metrics they are not compensated on, and that truth is often the most valuable data in the business.

Fewer, and older

Organizations accumulate metrics. Each was added for a reason and none are ever removed, so leaders end up reviewing forty numbers, none of them seriously. Five or six that stay stable for years produce better decisions than thirty refreshed quarterly, because stability is what lets you see trend and learn what actually moves them.

When you add one, remove one. It forces the argument about what matters.

Read the qualitative alongside it

Numbers tell you that something changed. They rarely tell you why, and the why is where the decision lives. Pair every quantitative review with a small amount of direct contact: read ten support tickets, sit in on two customer calls, talk to three people doing the work.

Leaders who do this catch distortions early, because the gap between the improving number and what they are hearing is obvious long before it appears in the data.