Speaking as someone who's been working on NLP, sentiment and semantic analysis for two years now, I've come to the conclusion that sentiment as a product is use

Speaking as someone who's been working on NLP, sentiment and semantic analysis for two years now, I've come to the conclusion that sentiment as a product is useless. There's no actionable conclusion from sentiment alone. You need to dig deeper, apply part of speech tagging, and determine the underlying intent. We, too, use SVMs to determine sentiment (in fact, we built a web-based API to create, train and deploy your own SVMs) but if you're goal is to make money, then you need more than just "this tweet is positive/negative/neutral". Marketing folks need to sell an actionable strategy. For example, "People are speaking positively about your product because they love the packaging" or "People are speaking negatively about BlackBerry because they can't figure out how to configure their work email accounts". Both of these give much more insight than just polarity.