It is better not to pre-focus on concrete metrics too early. Better have the mindset of a scientist trying to figure out an unknown phenomenon. Observe, look fo

It is better not to pre-focus on concrete metrics too early. Better have the mindset of a scientist trying to figure out an unknown phenomenon. Observe, look for patterns, ask the what-if questions, form hypotheses, try to verify them, etc. Not my direct experience, I just once helped a guy who started with this approach and needed some better scripts to gather and analyze more data. This is fun stuff, everybody should try it.