Marketers I talk to are hungry for advice on how to optimize social media networks, and so I’m happy to share some intriguing Twitter data recently collected by researchers at Carnegie Mellon University, MIT and Georgia Tech.
For this new study, CMU post-doctoral fellow Paul André, in conjunction with doctoral students Michael Bernstein (MIT) and Kurt Luther (GT) set up a website with the provocative name “Who Gives a Tweet?” They invited people to anonymously evaluate Twitter messages in return for having their own tweets evaluated by others. In all, more than 1,400 site visitors rated nearly 44,000 tweets over a two-and-a-half week period.
The results were revealing. Despite the astounding popularity of the Twitter platform, study participants liked only 36 percent of all the Tweets they reviewed. They disliked 25 percent of them, and nearly 40 percent of the Tweets evoked no opinion either way. If you add together those last two categories, you’re left with this sobering thought: A full 65 percent of posts on Twitter provide little or no value to people who read them.
So, what kind of Tweets did people like? The data showed that participants in this study liked three different types of Tweets: 1) those that shared information or provided links to good information, 2) those that were posed as questions and 3) Tweets that were self-promotional in nature, such as bloggers letting people know about a new post, or someone sharing an accomplishment.
What kind of Tweets turned people off? Links to old news (after all, Twitter is all about real-time communication), dull personal details (such as what someone was eating or where they were Tweeting from), griping/complaining, long Tweets that someone could not easily reTweet and messages that seemed too personal for such a public place as Twitter.
The three researchers who put this study together admitted that these results may be skewed towards the interests of the study participants, most of whom came from a techy perspective, and so may value sharing information more than people in others groups. As Luther put it, “Other groups within Twitter may value different types of tweets for entirely different reasons.”
Still, data such as this could assist with the development of apps to screen out certain types of disliked tweets or display messages in ways specified by a particular user’s preferences. And certainly, it can help marketing professionals use Twitter more effectively, crafting the kind of social media messaging consumers would be most likely to read, respond to and share.