Finding the real problem first.

How I frame problems, talk to users, test assumptions and decide what is worth building.