You could say it more formally as, "Many variables correlated to outcomes you want are not causes of those outcomes. When you intervene and change those variables, they cease to be correlated to the outcomes," but it's a bit verbose[0]. I've always read it to be not about measuring things, but about optimizing the wrong thing, because people will game the system. The concept is valid. What's your concern?

[0] https://en.wikipedia.org/wiki/Goodhart%27s_law#Expressions Turns out the original formulation was "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes."