Facebook can do test rollouts of features to specific segments of their users and exactly measure the impact. In comparison, making a survey gives you crap data that tells you nothing about how the usage patterns of Facebook will change. Because what people say they will do and what they actually do are two different things, and Facebook keeps very quiet about the actual patterns.
Sometimes though you can infer that announced features actually wasn't a success. They'll just be quietly rolled back or changed or never rolled out to the entire userbase. What happened to Timeline, for example?
If the interface is confusing, then it sounds like those extra billion image views a day were probably accidental -- which would simply mean wasted bandwidth (and money) for Facebook. Unless those extra image views somehow also produced additional ad impressions for Facebook or something like that, then I'm not sure why they'd consider having to serve up more images a success if people were happier using the old system...
I have no doubt that Facebook extensively tests their changes to the UI features with a focus on how much more time people spend on Facebook. For example, the lightbox photo UI was panned as being confusing and slow, but brings in an extra billion (+5%) photo views a day [1].
While 48% of teens claim that they try to ignore or see no real value in the ticker, I bet that a large chunk of them are clicking on it (whether they consciously want to or not).
[1]: https://www.facebook.com/notes/facebook-engineering/developi...