Routes do not fail suddenly. They drift, and drift is only visible against a baseline.
Why it matters
Every individual exception is defensible. The customer really did need that date; the technician really was the only one available. Added up over a quarter, the same reasonable decisions produce a route book nobody designed and nobody can defend. A health read catches the slope while it is still cheap to correct.
How you notice it slipping
Routing is only discussed when something breaks. The last real look at the routes was during an implementation. Nobody can say whether miles per stop is better or worse than six months ago, which means the answer is almost certainly worse.
What to do
The point of a health read is not to produce a report. It is to make the drift arguable — to turn “dispatch feels harder lately” into a line on a page that somebody can disagree with.
Pick four numbers and read them weekly: miles per completed stop, finish variance, exceptions after the lock, and the share of recurring appointments that moved. Four numbers, one page, every week. The value is not in any single reading; it is in noticing the week the line turns.