For all their analytics, you'd think they'd have figured out that I'm really loathe to sign up for an account when there's absolutely no indication how much it will cost to use it on my site. :)
This kind of data has a different kind of selection bias.
When you're looking at user actions, you can't discern two different kinds of behavior: (1) The user is doing what he wants to do; versus (2) the user is doing something because he can't figure out the alternative, or that alternative is inconvenient/unwieldy.
So to a certain extent, the results of such studies can server to entrench bad design decisions, rather than improve the system. The developers always have to look for every possible alternative explanation for a user's behavior.
That's very true - analytics cannot entirely replace talking to you customers. We still do that all the time.
They supplement each other quite well, though. Talking to your customers can help negate the form of selection bias you bring up, and looking at aggregate data can help you figure out if the needs of the most vocal 1% are worth pursuing right now.
