“What matters” should determine what gets “measured” - not the other way round. But availability of evidence may constrain a practical set of criteria. How can we navigate this?