When a Measure Becomes a Target

In 1975, the economist Charles Goodhart, who was advising the Bank of England, wrote that any observed statistical regularity tends to collapse once pressure is placed on it for control purposes.

Two decades later, the anthropologist Marilyn Strathern gave it the version that travels: when a measure becomes a target, it ceases to be a good measure.

Semiotics gives that warning a structure. The philosopher Charles Peirce described a sign as three things working together: the sign itself, the thing it stands for, and the interpreter who connects them. A dashboard number is the sign. Customer health, revenue quality, or safety is the thing. A manager is the interpreter.

A metric is trustworthy only while all three connections hold. Once the number becomes a target, people stop interpreting it and start serving it. The sign stays on the screen, still precise to two decimals, while the thing it stood for has quietly moved somewhere else.

AI sharpens the problem, because a model doesn't interpret anything. It optimizes. It will keep the sign intact and it cannot tell you when the thing has left.

So the useful question for a leader is which of the three connections is currently the weakest: the number to the thing, the thing to the goal, or the person who is supposed to be interpreting.

Which of your metrics has been a target the longest?