The truth is, customers talk about everything on the social web, and that includes talking about and directly with the brands that they buy from or have heard about. Some of this conversation has meaningful sentiment: posts like: "Thank you, I love what you did!" are certainly positive while: "This sucks, once again it doesn't work. #FAIL" is obviously negative. Look past the obvious, though, and sentiment fades as an indicator of what matters on the social web.
Given all of that, what's the real value of sentiment analysis? At best, it's a trendable measure of brand satisfaction; though, again, being "happy" and being "highly likely to recommend" are two different things. To be sure, an overall "positive" conversation is a good thing, and so measuring happiness makes a certain amount of sense. But I still have this uneasy feeling: when the conversation is upbeat, is it because everyone is happy, or because, like most of us, people are prone to being nice? Is my happy index going to withstand an ice-storm-driven outage that disrupts my customer deliveries for a week?