Thread by @johncutlefish

Typically when someone asks "What should we measure?", my response is (something like) What decisions do you need to inform? Where do you need to reduce uncertainty? What do you need to learn?

This helps with *some* ppl, but not with others..(1/6) #proddev @Amplitude_HQ

...I've been mulling this over. 1) for many ppl, metrics are like "grades". They are like students asking "how do you get an A?", "are you grading on the curve?" So...not about learning, and more about judgement and pass/fail. Problematic ..(2/6) #proddev @Amplitude_HQ

There is an intense worry that the team will pick the "wrong" metrics, and bad things will happen. Which in some environments is a distinct possibility. What this misses is that measurement itself is often an iterative process. (3/6) #proddev @Amplitude_HQ

They may have been in an environment where only "what was measured" got managed (an oversimplification of Deming blog.deming.org/2015/08/myth-i… ). In which case they are right to be especially paranoid of "picking the wrong things to measure". (4/6) #proddev @Amplitude_HQ

For many teams it is extremely, extremely difficult to talk about their bets, uncertainty, risk, etc. That's not a muscle they use often for whatever reason. In that situation, at best you get a vague "to confirm we're taking the right direction" (5/6) #proddev @Amplitude_HQ

Start with: What decisions do we need to inform? Where do we need to reduce uncertainty? What do we need to learn? Converge on that, and then the measurement challenge becomes much, much simpler (6/6) #proddev @Amplitude_HQ