I'm an experienced user with Analytics (coming to 20 years already) with a programming background, and it was in the last 8 years or so that I've found AA's reporting relatively easy to comprehend when I imagined it as a series of SQL queries.
Having said that, the 2 biggest hard-to-use components of AA are segmentation and fallout/flow visualisations. Not that their UIs are difficult to use, but the resulting data/reports are not easy to understand.
For segmentation, I've found that drawing block diagrams (e.g.a. square is a "hit", a row of squares is a "visit", a collection of rows of squares is a "visitor") helps to get me closer to understanding a segment's results, but not always. The documentation doesn't really help any better in that understanding.
Related to segmentation are the fallout/flow visualisations. At a conceptual level, I understand how the numbers are derived. But even when I look at the segments used to build the visualisations, I find myself having to read and re-read the segment conditions to really figure out what is going on.
It is said that if you can describe or teach something to someone else, then you've mastered that something. For these 2 components, I still feel that I'm not ready to teach others sufficiently about them.