A few years ago a company named Klout appeared that promised to be able to measure social media influence using what they called the Klout score. Today Klout was bought out by some marketing automation company called Lithium Tech.
When I hear of Klout, I was so surprised that someone had finally and quickly solved the never-ending challenge of measuring and ranking social influence through a mathematical algorithm, so I jumped right into it, created an account and signed up to have my score measured.
Only to find out that the scores were nonsensical, the automated topic-choosing algorithm was a little (or a lot) off and that they still had a long way to go. I won’t pretend like I know what algorithm Klout was using to measure or quantify influence, but I had to guess that it had something to do with Social Network analysis, an area that I’ve done quite a bit of practical work in. the idea behind SNA is that one should be able to learn about the influence of a person from their position in any given social network. The basic measures we use to identify position are things like Eigenvector centrality, betweenness centrality, and degree centrality–essentially how central a person is to a follow/friend network, or how central one is to a given conversation (@ replies and re-tweets for a network like twitter).
Klout got a few or more things right. For example, their use of an absolute score from 0-100 which was clearly non-linear was a great start. social experience and influence in particular are almost always non-linear. In other words, just because I saw 5 of your tweets today, and 1 of someone else’s does not mean that I was influenced 5 times more by you than someone else. But, the system itself created some oxymoron: Knowing that Justin Bieber has a higher Klout score than the pope does that mean that he is more influential than his holiness?
Klout’s answer: No. They are influential about different topics!
Then how do we define topics, and wouldn’t having to rely so heavily on Topic Influence as part of your algorithm mean that your algorithm must be superb at defining topics of influence? This is where Klout falls apart and why they never really did succeed at finally mapping social influence. First, it doesn’t take a genius to sign up to Klout and notice that the topics that were selected for his/her account do not match their conversations. Second, there may be instances where Justin Bieber is influential about a topic and his influence is directly contradicting what the pope might say about the topic, but because the Pope’s offline influence is likely much broader and stronger than Bieber, does that not affect his ability to influence online?
These types of ambiguous and difficult to answer questions is why Klout did not make it, but they did do a great job of getting us closer to an answer.
Perhaps one must look at online influence with a much broader lens. Maybe influence is not topical online as it is in real life. Yes, I won’t take financial advice in real life from, say, a baseball coach, because in my mind that is not an authority on that subject, but I would take advice on say, sports, from the same person. Online, “advice” would not necessarily be my objective in the first place, so topical influence may not be at play to start with.
MY data collection during the last presidential election showed me two things with absolute certainty. The majority of “influencees” on social media craved two things: News, and Entertainment–not advice. This is where Klout would’ve had their own Billion dollar IPO. If they targeted the measurement and delivery of tools to improve the influence and trustworthiness of any new media blogger, or old school news outlet, and helped those in entertainment build better marketing campaigns.
These things are not topical!!! If you want to get the latest news about something you go to where the news you want is located and then you gran all the news you want from that outlet. Some people have 3 or 4 sources for their news, but they generally do not, say, go to CNN for news about the Treasury, and then to CNBC for news about the FED–they usually go to, say, CNBC for news about Business–ALL business. Why? because they want a brand they can trust, regardless of specific topic. Plus its easier to define if a topic falls into the category “Business” than it is to define that it falls specifically into “M&A”s, which makes topical recognition easier for the algorithm.
Fear not though, Klout will still be around under Lithium Tech., though instead of measuring the World’s influence they shall be relegated to the gates of an ambiguous “social customer experience solutions” which is a fancy description for marketing automation for big name brands.
Good Bye Klout! You really did give us hope for a while!