A two-week audit produces a lot of findings. Exactly one of them matters enough to put on a slide, and choosing it is most of the value.
The number has to survive three tests: someone owns it, it moves within a quarter, and a sceptic can check it without trusting us.
Median first-reply time passes all three. “Customer satisfaction” passes none of them, which is why it appears in so many decks and so few decisions.