RankSpeed, an early player in searching Twitter and blog content for how people feel about subjects, relaunched itself today as a place where you can learn about products based on sentiment.
For example, you can search for ‘Facebook apps’ and the word ‘excellent’ and see which products are mentioned the most with that word. RankSpeed will show the percentage of users who have talked about the app and used words like ‘useful’ or ‘terrible’ along with it, in time periods ranging from one month to all time. If the company helps drive sales of products, RankSpeed could earn affiliate fees for referring customers to sites like Amazon.com.
Several companies are trying to make sense of the opinions expressed en masse through status updates on social networks like Twitter and Facebook, but we have yet to see a breakout hit. RankSpeed is promising, but it also illustrates some of sentiment analysis’ current limits. For example, if you search for ‘Facebook apps’ and ‘excellent,’ group buying company LivingSocial comes up in fifth place. But if you search for ‘Facebook apps’ and ‘terrible,’ the company comes up in second place. Many products elicit deeply divided reactions, like Twitter for example, and RankSpeed isn’t good at accommodating that. It also doesn’t replace the experience of researching a product thoroughly; if I was going to spend $1,000 or more on a product, I would read many in-depth reviews and I would ask friends. RankSpeed might be an accessory to this process, but it wouldn’t be the ultimate tool I would use.
VB's research team is studying web-personalization... Chime in here, and we’ll share the results.