The short answer is: a lot. If your social media efforts are strictly tied to tweeting out coupon codes, then you have a pretty direct and reliable measure of the efficacy of your efforts: simple conversion. For the rest of us (and, that's pretty much 99% of us), online measures are not going to capture the full impact and value of social media for your organization. There are lots of smart folks in social media monitoring and sentiment analysis trying to crack the ROI nut, trust me - but mining unstructured data alone will never truly quantify the value of online engagement to offline sales. For instance, Southwest Airlines (SWA) has a notable presence in social media, particularly with their Twitter account at @southwestair. A lot of people talk about them online, and a lot of the chatter about SWA (particularly in comparison to some of their U.S. competitors) is positive. Buying air travel, however, is not a spur-of-the-moment decision for most people (myself excluded :)) There is likely to be considerable distance between the point of influence - being favorably predisposed to SWA through their social media interaction - and the point of purchase. In the intervening time, SWA is also likely to run some kind of sale or promotion, and while it may have been that promotion that actually prompted the purchase, it may also have been Southwest's online behavior that put them into the purchaser's consideration set.
See where I am going with this? If the offline sale gets the "credit" for conversion, the efforts expended in social media - however important - get little or none. Parsing out the impact of cross-channel media on purchase behavior is a bit of rocket surgery, but well within the purview of a competent CMO - provided, of course, the right inputs are available. If you are only measuring your social media efforts by mining unstructured online data (monitoring, sentiment analysis, etc), then you may be capturing enough to track reputation, or the health of your brand on the social web, but you aren't tracking enough to make the connection to purchase behavior, particularly in longer sales cycles.
This isn't to say that you shouldn't be monitoring or mining this data online; it's an essential input to the process. You do, however, need to augment it with offline inquiries. This may mean commissioning additional online or offline research projects (depending on where your transactions actually happen), or perhaps adding a few social media indicators into the data you already collect, but the bottom line is that this isn't really a mystery. It's done every day by marketing departments all over the world, for other channels and media.
Here's what prompted me to go down this particular rathole today: I am currently working on a project to measure the impact of a campaign that will target visitors to minor league ballparks across America. The campaign, which has a variety of components, hasn't run yet, but we are measuring today, in the actual ballparks, before a single tweet, billboard or radio ad has begun. By conducting offline pre-engagement measures at those ballparks, we can sample the fans and develop a reliable baseline for our client. Later, after the campaign has run, we'll conduct the exact same measures, in the exact same way. Our client will know precisely what worked and what didn't work (online and offline), because we will have the apples-to-apples, before-and-after analysis to determine what components were successful, and what aspects didn't perform.
Pretty easy stuff, really: measure before, measure during, and measure after. Measure online, but also measure offline. Works for out-of-home media, works for TV and it will work just fine to quantify the value of your Twitter account. Mining online data can give you a snapshot of what people think about your brand or product on the social web, and tracking this data might even give you a sense of how these perceptions change over time. For most brands, however, the actual behavior change occurs elsewhere. It might happen in an Amazon shopping cart, at a car dealer, or even at a local ballpark. When you can sync your online monitoring efforts with offline measures, one calibrates the other - and the true ROI of social media can be measured, understood and appreciated. For my friends in the social media space, that is what I want for you.