Working on analytics for an online application based in a social media platform has been interesting, from a couple of angles.
1. measurement of virality of the application.
You have your plan of how your user-base will grow, but only through analysis of user data will you learn what is really happening.
2. focus on user engagement
Online retail for instance wants you to buy something; look around the store, pick up a couple of things and go through the checkout. Social? Well there could be any revenue model, but what you really want is users interacting with the app, with each other, and with the thing that is generating your revenue, no clear funnel necessarily, a lot of chaos to make sense of.
I would not go so far as to say that Analytics for a retail site vs social is totally different, but the focus is slightly shifted. eg on a retail site you may ramp up your ad spend as you grow, where on social you want to get a critical mass early on and gain a groundswell of users inviting their friends as soon in the application’s life as possible.
An article on ClickZ (sponsored by Omniture) Web Analytics for Social Media starts to look at this issue, but leaves it with a lot of open questions, but with social media being such a broad area it would be tough to go into specifics.