as measuring code coverage for a team doing TDD.
The process/ways of working get you results so good that you don’t need to measure them at all.
It’s valuable to ask yourself if the need for metrics you’re measuring is actually an accidental complexity of the underlying suboptimal ways of working.
Once you fix that, the need for measurements, infrastructure to collect them, products you pay to get the results, all the meetings and retrospectives around the measurements, etc. is usually gone.