Your Followers Are No Measure of Your Influence
One of the nasty side effects of the rapid growth of social media is that it threatens to warp our understanding of influence. It's only natural that Twitter has given rise to any number of applications that rank users for various criteria, including their overall influence. Many of the 150 million or so of Twitter accounts contain multitudes: a feed of interests, passions and expertise, in many cases attached to a living, breathing, identifiable human whose popularity is neatly summed up by follower counts, the lists he or she is on, and the number of times he or she has been retweeted. But a marketer has to wonder what all that information means, if it adds up to anything more than a popularity contest and what, exactly, does a tweet influence a person to think, believe or do?
It's hard to imagine that Justin Bieber, with his 6.4 million followers, is driving much behavior other than getting people to talk about Justin Bieber, frenetically retweet him, and possibly buy a record. Is that influence?
Klout, an online service that describes itself as nothing less than the "standard of influence," thinks so. Its algorithm gives Mr. Bieber a perfect score of 100. "You can't get any more influential than this," reads his summary. "People hang on your every word, and share your content like no other. You're probably famous in real life and your fans simply can't get enough."
Indeed, Mr. Bieber is famous and, as a YouTube discovery, his fame has been built on social media. He has a prodigious understanding of how to use these tools that helped him rise out of Canadian obscurity before he had to shave. Yet it's hard to imagine how he is a paragon of influence. He simply seems popular.
Some recent research by Duncan Watts and three other researchers shows the problems with popularity. Mr. Watts, now a researcher for Yahoo, caused a stir a few years back with work that challenged the validity of "Tipping Point"-style thinking about the way influence works. Equipped with evidence that showed cascades -- chain reactions where one user passes something to another, and so on -- are nearly impossibly to predict, he argued that, rather than focus on finding a few, highly influential people to spread a message, anyone who wants to "go viral" should be on getting a message to as many people as possible. In other words, you have to hedge your bets and not simply rely on your models of influence, however finely honed they might be.
In a 2009 experiment in Twitter, Mr. Watts found that those findings were transferable to the then 3-year-old microblog. He and fellow Yahoo-er Mr. Mason looked at more than 1.6 million users and 74 million instances of sharing of something, known as "diffusion events." In many cases, the most popular Tweeters generated the biggest cascades. That's no surprise. "However," Messrs. Watts and Mason wrote, "we find that predictions of which particular user or URL will generate large cascades are relatively unreliable. We conclude, therefore, that word-of-mouth diffusion can only be harnessed reliably by targeting large numbers of potential influencers, thereby capturing average effects." In other words, reaching a large number of more ordinary Joes and Janes with a message might be more effective than trying to tap into Bieber fever.
When people try to think past follower count to a more nuanced metric, they might end up with something like what Twinfluence, a rival to Klout, describes as social capital. That metric combines the influence of a tweeter's followers with his followers' followers. Or they end up with something like Klout's amplification metric, which charts the likelihood that a tweet will spark some action. All this sounds nice, except for the fact there is only so much you can do with a tweet. You can retweet it or you can make the tweets one of your favorites or you can use the tweet as a stepping-off point for a conversation. And that's about it.
For those of us in the content game, that's fairly useful. The same goes for marketers who want to talk to their customers or give the appearance that they talk to their customers. For parties with other kinds of goal, that utility is less clear.
Think of Twitter as a game with just a few objectives: earn followers and retweets and clicks on your links. While services like Klout are wonderful at judging the winners on those rules, they're not as good -- even useless -- at providing a means of understanding how that particular performance might be extrapolated out to something as broad as influence. Thinking about this reminds me of studying for the SAT and coming across this bit in a Princeton Review book: "We're not big fans of the SAT. It doesn't measure intelligence. It can't possibly measure your future success in college. The SAT measures one thing, and one thing only: how good you are at taking the SAT."
The same might be said of many current ways of looking at effectiveness on Twitter: They have little respect for how an action on one of those networks might relate to behavior beyond Twitter.
Earlier this year, you may or may not have been swayed by Mr. Gladwell's controversial examination of the limitations of social media to formant consequential political activity. A similar argument can be made for marketing. There's a vast world of behavior beyond the retweet, from verbal word-of-mouth recommendations to actual purchases. Except in few cases, we struggle to monitor them. More than anything else, the limitations of a service like Klout might be a stand in for bigger problems in understanding how social media fits into the marketing's big picture.
I'm not totally sure why I came to remember the SAT line, but it may have to do with the fame-for-fame's sake quality of social media, the best/worst example of which can be found in a marketing stunt from the magazine Fast Company this summer. The Influence Project asked readers to create a profile with a unique URL to be shared by as many people as possible. The winner would go on the cover of the magazine as the most influential person on the web.
The project sparked both an enormous amount of tweeting and Facebooking and a rather nasty backlash. One newspaper dubbed it a "botched social media campaign." A few folks plotted to hijack it. One blogger asked, "What happened to the days when having influence meant producing thought provoking ideas and reactions?" Another, Amber Naslund, VP at the social-media monitoring firm Radian6, wrote: "To me, influence isn't about popularity. Or even reach. It's about the trust, authority, and presence to drive relevant actions within your community that create something of substance."
The final argument against looking at Twitter as a de facto measure of influence is so steeped in common sense, it might offend the intelligence: Simply look at who doesn't spend much time there. One of them is Seth Godin, by any measure, including Ad Age's Power 150, a thought leader for marketers and entrepreneurs and a popular blogger who uses Twitter (and Facebook) only for rote repostings on his blog posts. Celebrities like Kanye West are routinely late to the game and don't seem to suffer much for it.
And then there's Apple. Every marketer's favorite brand to this day still doesn't give a Nano about using Twitter proactively. There was a brief flicker of excitement in July when Scott Forstall, the senior VP in charge of Apple's mobile operating system, signed on and, rare for a less-than-household-name business executive, received a verified account. Today, Klout gives Mr. Forstall a score of 59 and credits him with "high percentage of amplified content." Thirty-six thousand people follow him, while he follows just one, Conan O' Brien. Guess how many tweets the appropriately-named Mr. Forstall has posted.
Not a single one.
How's that for influence?
|ABOUT THE AUTHOR|
Matthew Creamer is an editor at large for Advertising Age.