The bit in red is mobile; above that is desktop; below that is tablet. The tablet audience is surprisingly small; it's swamped by both desktop and phone. The "bounce" rate is the percentage of people who hit the site and only look at one page. The pages per visit is a mean, and includes the bouncers. The time-on-site thing I consider kind of unreliable, to be honest. (It's regularly internally contradictory within a single day, is why.)
Anyway, as you can see, phone users used to bounce off the site at a dramatically higher rate than other kinds of users; now they've moved into line with everyone else. Pages per visit on phone is up, time on site (unreliable tho' it may be) is also meaningfully up. Mobile users had been dragging the whole site average down; now it's back in line with other formats.
So I guess that mobile view is kind of important! In retrospect I guess that seems kind of obvious, but I tend to hate them personally, so I guess that kind of affected my impression of their value. I wish I'd done the work earlier, now. Ah, well, live and learn.
Echoed from Crime and the Blog of Evil