Perhaps a quarter of total web traffic occurs on social networks and that number could double if the share was restricted to non-corpaorate or academic use. This number, however, does not convert in anyway to the share of total web advertising expenditure. In contrast the social newtorks are the main source of the lowest cost form of inventory that provide the impression bank for the myriad ad networks that add value through targeting and re-selling to advertisers mostly on a performance basis.
The outcome for the social networks is not a happy one. They accrue limited revenue and open their environments to a volume of advertising of varying degrees of relevance little of which is welcomed by their users.
Facebook is of course not just one network but a central set of tools and utilities on which many separate and interlocking networks operate. The Harvard Alumni, the Obama supporters, the WPP employees and so it goes. Given that this data is self reported and (unusually for such data) mostly accurate does this set up a more relevant ad model?
The operation would involve Facebook selling access to the their users and the networks they create by allowing, for example, advertisers to target the Harvard group, not on Facebook, but elsewhere on the web as they use other sites who also sell their inventory to the network operators.
In so doing Facebook becomes a fascinating behavioral targeting play in its own right without compromising its own properties. The question this raises is one of privacy, will Facebook users object to their data being used in this way? I suspect not; the data are not personally identifiable and its users are already accustomed to the absence of free lunches in media consumption.