"For a guy who moved all day long, Paulie didn't talk to six people."
-- Goodfellas (1990)
"Influence" is making headlines again as marketers seek to shape the conversations relevant to their organizations across the social web. Most recently Chevy partnered with Klout to identify influencers for a promotion aimed at driving positive word-of-mouth buzz around its Sonic line of cars. Chevy is offering three-day test-drives to Klout users with scores over 35. However, what if Klout score's aren't an accurate measure of influence?
The most commonly held belief about influence is that it is equivalent to the number of a user's friends. Klout, for example, measures influence using three different sets of metrics but the underlying assumption of all of them is a direct correlation between influence and number of fans. The problem with this model is that it may not be correct. A recent study published in the May 12th issue of the scientific journal, Nature, suggests that the number of friends a person has may not provide a true measure of that person's influence across the social web.
In the Nature study (conducted by researchers from Northeastern University, the Dana-Farber Cancer Institute, Massachusetts Institute of Technology and Harvard Medical School), the researchers used principles of network science and tools from the domain of control theory to analyze a number of real networks, including social networks. They identified the "driver nodes" of these networks - the points into which they could inject control inputs to direct the networks according to their wishes - and made a very surprising discovery. The influential connecting points in these networks - the "driver nodes" - tended not to be the network "hubs" or the ones with the most connections.
In other words, the members of the social networks with the most friends were not the ones best ones for influencing the behavior of the networks.
This discovery is not as counter-intuitive as it sounds. I'm sure all of us are familiar with influential and powerful people who actually interact with very few people on a given day. The most obvious example is Steve Jobs. Here's someone who wasn't on any of the major social networks and yet every social network rippled whenever he uttered a syllable.
Perhaps Jobs is not the best example as his presence in the general culture made him easy to identify. This is not the case with social media influencers. Most of these people are only influential in specific niches across the social web and are thus difficult to identify for outsiders. However we can trace their influence using listening tools. By listening to the conversations we care about and tracing the dominant memes and opinions to their sources we can successfully identify the "driver nodes" of the social networks that are important to our businesses and organizations.
Certainly, the last word on measuring influence and identifying influencers in social media has not been written, despite the fascinating findings of one recent study. However marketers should be aware that there's no easy answer to this conundrum, as much as we would like to use friend and fan numbers as surrogate gauges of influence. Most importantly, it's important that marketers seeking to shape the conversations they care about understand that identifying influence requires careful listening, rigorous analysis and, in short, hard work.