Yeah that's the thing. I was talking about computed columns in tables, materialized (or indexed) views, about measuring stuff by experiment and about measuring stuff that actually matters but the interviewer seemed like he wanted to hear a simple answer