Senti-mentalism

One of the more thought-provoking blog posts I’ve read recently is “Five Reasons Sentiment Analysis Won’t Ever Be Enough” from Marta Strickland on the Threeminds blog (linked below). Not just for the specifics of the post, but for its reminder that a lot of what gets called “analysis” in social media reallyonly goes halfway.

Social media people, in my experience, are good at sourcing data. They’re often good at crunching out the patterns in it. And increasingly they’re great at working out how to display it. But that’s not the limits of “analysis”. You need to know what the information means, which in a commercial research context means knowing what the people buying it want it for. “Find out what consumers are saying about your brand!” Great! Then what?

(I’m not saying that researchers are paragons of virtue when it comes to answering “then what?” – one of the criticisms that’s often been aimed at our industry is that we often fall in love with information for its own sake, beautiful or fascinating data over useful data. And no surprise: if we didn’t find people interesting and want to ask them things, we wouldn’t be in the business.)

So as sentiment analysis gets better and better – and it will – we should always remind ourselves that information is not interesting in its own right, no matter how consumer-generated or ‘organic’ it is.