Twitter currently hosts more than 215 million monthly active users, who generate more than 500 million tweets per day. Last month, the social media platform went public, raising $1.82 billion. Those are big numbers for a seven-year-old startup, but what do they mean for CIOs and the businesses they work for?
Well, if tweets are a window into customer behavior, and understanding customer behavior keeps businesses relevant, tweet analysis has potential business value -- and therefore IT value.
The key word is potential. For all the talk about analyzing social media for business gain, Twitter -- and social media in general -- remains rugged terrain, where machines can spew out human-like utterances, data can be dirty, and language can be misunderstood and exaggerated. The data-quality challenge alone is a call to arms for IT organizations, and there are many more aspects of Twitter analysis that cry out for IT expertise.
Nowhere is the need for IT expertise in social media analysis more evident than in some of the cutting-edge research in the field.
Social media in general remains rugged terrain, where machines can spew out human-like utterances, data can be dirty, and language can be misunderstood and exaggerated.
Sherry Emery is a senior research scientist at the University of Illinois at Chicago's Institute for Health Research and Policy and an investigator of tobacco industry marketing. She has turned to Twitter to study the subculture of smoking and how the use of tobacco-related products evolves. The unfiltered chatter on tobacco, marijuana and electronic cigarettes offers another window into what's happening with these products.
To that end, her team has collected more than 400,000 tweets on e-cigarettes alone in the past year. It's an emerging market, one that doesn't carry the same social stigma and isn't subjected to the same FDA regulations as traditional cigarettes, she said. Emery is interested in deciphering who is doing the social media talking: Are the tweets coming from spammers, marketers, bots or enthusiastic users?
"We see lots of evidence of spam marketing," Emery said. "It suggests to me that there's a ton of money behind the promotion of these products."
But differentiating between actor and consumer and, therefore, between what someone is paid to say versus what someone really thinks is tricky. Messages about e-cigarettes tend not to be brand specific, and many appear to come from offshore marketers, making it difficult to track down a connection between tweeter and e-cig vendor, Emery said. Tweets can reveal location, but location services are an opt-in service that tweeters rarely turn on, used in only 5% of the data.
Tweet analysis yields unfiltered chatter on what teens are smoking
Despite the flaws and imperfections, the value contained within tweets should not be overlooked, Emery said. Her team is able to tap into segments of the population that are underrepresented in traditional surveys and keep tabs on how new products, such as e-cigarettes, are trending.
Carol Haney, who works with the Centers for Disease Control (CDC) and is the senior vice president for group product marketing at London-based market researcher Toluna, agrees. She is measuring a new market for cigarillos and clove cigarettes among teens and 20-somethings. Sliced open, emptied of tobacco and refilled with marijuana, the cigarillos and clove cigarettes are refashioned as blunts, street talk for a kind of hybrid smoke.
"Cloves and cigarillos have highly concentrated fruit juice in them, and that flavor, some think, masks the smell of marijuana," Haney said at the Text Analytics Summit this past summer. "It's increasing [the popularity] of a type of … cigarette that otherwise was having really low market share." The trend is concerning because, in addition to concentrated fruit juice, trace amounts of nicotine and tobacco cling to the wrappings used for blunts and can lead to addiction, she said.
Don't take emoticons at face value
☹ Toluna's Carol Haney studied emoticons in tweets that discuss smoking. She learned that in the United States, the frowny face is a good predictor for tweets containing negative sentiment and substantive content. "If someone uses a frowny face, they're actually saying something of interest for formative reasons because they're saying something about how they're feeling," Haney said.
☺The smiley face, on the other hand, is not necessarily a good predictor of positive sentiment because, while it's often used in connection with humor, it's also used to convey sarcasm.
Defining this market is a many-layered task. Haney studies not only the text from each post she gathers, but also the use of hashtags, images, retweets and URLs. She "listens" to what is being said, and she searches for patterns to figure out how tweets are constructed. One hypothesis: When URLs and hashtags are used together in tweets on smoking, it's a signal for blunts.
To test her theory, Haney pulled tweets that contained such words as cigarillo, Swisher (a kind of cigarillo connected to the use of blunts) and so on. In 72% of the tweets she studied, the hashtag Swisher co-occurred with a URL. The insight on how a tweet is constructed can be vitally important when trying to communicate with that population of users.
"It's fascinating, because there's a whole lingo around how people relate to using those cigarettes," she said. "And if you're running a counter-marketing campaign, you'll want to know that so you can use that same language … to redirect people out of using tobacco cigarettes for smoking marijuana, because it also gets you hooked on the tobacco."
Of course, there's a lesson in all this for businesses and health organizations besides those connected with tobacco addiction, points out Jenny Sussin, an analyst at the Stamford, Conn.-based consultancy Gartner Inc. "You see a lot of market research and marketing campaigns being built [around Tweet talk] for the same reason," she said. "Businesses want to make sure they're communicating on the plane [customers] want to communicate on."
Read part two of this story to learn more about using tweet analysis to communicate in the lingo of the customer and about Gartner's research on how IT involvement drives social analytics success.