That's an interesting question raised by Andrew Smith of Escherman, a communications shop in the UK. Check out his blog, How to guarantee a successful career in PR for $30, which is a review of Avinash Kaushik's book, Web Analytics: An Hour a Day. Smith has provided a great overview, notwithstanding that the book is available at Amazon for less than $20 (excluding shipping).
Here's his summary about metrics, which I think is especially important in looking at social media/Web 2.0 campaigns:
"While the industry still seems to be floundering around trying to develop an acceptable standard for PR evaluation, the Web analytics industry can now potentially offer the ability to connect PR value to real business outcomes. There is no reason why PR campaigns can’t now be built that can be measured and evaluated in the context of metrics that really matter to a business rather than busted flush approaches like advertising equivalence."
Kaushik also provides the 10/90 rule, which says firms ought to allocate 10% of their budgets for tools and the rest to paying for human beings with analytical skills. The challenge with that, I think, is that there are new tools to monitor the Web 2.0 world, and not all of them are free (as is suggested in the book, I believe). Data collection/monitoring can be hugely expensive or time consuming, which amounts to the same thing: the collection can be the tail that wags the dog; while the analysis is the important/essential part, compiling the data takes the most time.
I see this as getting worse as social networks, blogging, etc. continues to fragment the media and divides the mass audience into many, many smaller groups.
But that's the topic for another blog. I think that PR agencies and their clients should have metrics, and that those metrics need to be determined on a client- or project-basis because of the fragmenting of the audience. But the book, Web Analytics: An Hour a Day, seems like a good place to start.