I'll be one of the first to admit that formal ROI for many collaborative and socially-oriented technologies is a difficult task. e-Mail ROI efforts were always somewhat contrived back in the nineties as are efforts to "prove" ROI from instant message today. The more application-centric the implementation is however, the better the chance that you can define some set of cause-effect metrics related to a particular process. But overall, many aspects of "ROI" are qualitative. There are exceptions. Web conferencing for instance has real quantitative measures (reduced travel expenses).
But difficulty does not mean that the effort is a fool's exercise. Decision-makers need some set of criteria to determine how to best allocate scarce resources. Organizations only have so much discretionary resources (people, time, money) from which they allocate to support existing business activities and initiate new ones. So if ROI is an inappropriate method then some other assessment that can identify value-return is necessary. Some measures might look at skill/competency metrics, others measures might include additional items added to the supervisor/employee review process (e.g., team building). The issue comes back to one of the core missions of IT groups: helping business leaders and strategists select a subset of options from the total list of possible investments available. Some organizations alternatively apply "real option analysis" when traditional approaches are inadequate.
I've been getting a number of reporters asking about the ROI behind an application like Clearspace lately. My general response is that it's a fool's exercise. Trying to determine if the savings and revenue increase are worth the expense is like trying to measure whether the view from atop Everest was worth the climb -- it's exceedingly hard to measure and it should be painfully obvious.
Source: ROI, ShmoROI at Jive Talks