I'm running some ads on Facebook.
We're using a relatively focused, and therefore relatively expensive, list. Right now the "exposure meter" is clicking away, showing person after person seeing the ad, but no interest. Now remember this list is the same category of person that has shown strong interest in our work before. Suddenly, however, if this data is to be believed, there is no interest. Does it make sense that the list has suddenly changed its characteristics, or that maybe there is something wrong with the unaudited exposure data coming in?
The test will come when we compare social media to other outlets. We've had some success in certain print magazines before, and conference handouts, that have far exceeded social media results.
Update: we drastically changed the demographic focus of our Facebook ads to overcome what we perceived to be a potential bias error in their data, and now we're getting good results. In our opinion, Facebook data has some serious problems that need to be taken into account by anyone using them for advertising.