Trying to target ads to people based on their digital behavior is a hit-and-miss business, with the emphasis usually on miss. Tracking you browsing history too often results in ads that fall into the uncanny valley of internet advertising -- a Zappos banner that follows you all over the web, showing you the same pair of shoes you already bought and returned, for instance.
But monitoring your social media output doesn't yield much better results. If a marketer knows exactly what it's looking for -- a hashtag or keyword, say -- it's useful enough, but the applications of such passive monitoring don't go much beyond customer relationship management. To turn social chatter into actual sales leads, however, you have to be able to process a vast quantity of natural language and separate the tiny threads of signal from the oceans of static.
A three-year-old company called Viralheat does just that, crunching 2 billion social mentions a day to generate "predictive analytics" for customers like Deutsche Telekom and Men's Wearhouse. "Search is such a great conversion tool because people are typically typing in their intent," says Viralheat's CEO, Jeff Revoy. "Social is not such a great commerce tool because people aren't necessarily looking to buy something. But it is where they talk about things that are really relevant to fill out that profile of a customer."
Where social beats other platforms is in its ability to generate data that's prospective rather than retrospective. If you've gotten married, moved or had a baby in the last five years, chances are you announced it on Facebook (or Instagram or Pinterest) beforehand.
While marketers and the data miners they work with might be able to figure out the same thing in other ways, it comes with a significant creep factor. When I was shopping for an engagement ring last fall, I didn't really want ad for jewelry retailers popping up in every browser window where my girlfriend might see them, and I know the teenage girl who started getting coupons for pregnancy vitamins in the mail from Target didn't want anyone else to find out about her condition.
With social data, there are fewer such privacy concerns. "This is public data people are putting out there," says Revoy. "This is just giving companies better information so they don't use such blunt marketing tactics. If I know you're going to be a parent, there are very different ways I'd message to you than if I know you're single."
Recently, Viralheat was granted patents for two of its technologies, a method of natural language processing for sentiment analysis and lead identification software. The company has also partnered with Variety on a new index, called Vscore, that uses social media mentions to rank actors based on how the public feels about them. Variety intends for its subscribers to use Vscore, which monitors over 17,000 actors, as a tool for casting and other decisions, in much the same way agencies and brands use the better-known Q Scores.
