It’s good to measure learning/iterations — but those can just as well be vanity metrics as well. The question seems to be: how can we measure how close (and how much closer we’re getting to) predictable/repeatable impact?
That’s an interesting thought. Do organisations dedicate sufficient resources (time, etc.) to learning to make these types of partnership viable? I often see people engage in things for one reason at the beginning and then realise that the real value was something entirely different at the end — but they would never have started if you had tried to sell them on the final reason.