What gets measured gets warped

“How do we improve the train to Paris?”

“Cut the journey by forty minutes.”

“Anything else?”

“That is the improvement.”

The engineering answer is excellent once speed has been chosen as the definition of better. Billions can be spent, tracks rebuilt and success reported to the minute. The awkward question is whether passengers wanted shorter travel more than reliability, useful time, comfort or less anxiety.

The wine-and-supermodels example in the excerpt is deliberately daft. Its job is to make the hidden decision visible. If absurd luxury sounds like an answer, then “better” was never identical to “faster”.

Reliability complicates it further. Forty minutes saved on paper may mean little to someone who must allow an extra hour because delays vary wildly. A comfortable table and working connection could make the existing time useful. Neither improvement produces the heroic simplicity of a faster-train headline.

A support centre repeats the same mistake with average call length. Staff learn to end calls quickly. Repeat calls rise, but they belong to another column and another manager. Nothing fraudulent occurs. Everyone improves the figure attached to their name.

Parcels can arrive faster with more damage. Schools can lift tested scores by dropping material outside the exam. A metric can improve honestly while the thing people cared about gets worse.

At a review, one call-centre team reports thirty extra seconds per call, fewer repeat contacts and more problems solved first time.

“So duration is worse?”

“Yes.”

“And the service?”

Nobody answers immediately. After a pause, the chair moves to staffing.

Behavioural principles

Behavioural ideas at play in this post

Short, plain-English explanations of the principles behind this post, with links to related books and examples in the archive.