Companies such as PeerIndex and Klout claim to measure influence by weighing up combinations of platforms, followers, and sharing. Klout is currently aggressively promoting itself by pairing with brands; for example, Cathay Pacific is currently offering free entry to its SFO lounge to passengers with Klout scores above 40
, and Red Bull has worked with Klout in Australia
. It’s clear that brands are looking to partner with the most influential advocates available, but is Klout a reliable way to measure influence?
Klout scores start at 0 and Klout claims that the average user gets a score of 20 out of 100, but we decided to put that to some simple tests – in particular examining whether people who link it to their Facebook profile get a disproportionate lift.
Twitter-linked accounts: You’re a 10 even if you do nothing
To determine the baseline Klout score, we conducted the “Snow White and the Seven Dwarves” test. Seven Klout accounts were created named after the dwarves and appropriate content was tweeted daily (e.g. jokes from Happy). The dwarf Klout scores stayed at 10 for the entire month–even Sleepy who posted nothing and had no followers. This result suggests that baseline score is 10 and that the claimed average Klout score of 20 is very low.
Facebook-linked accounts: Is 40 the new 20?
Next we tracked the Klout of three male and four female undergraduates, aged between 18 and 28, who used Facebook, averaging around 400 friends and reporting only light Facebook activity. Astonishingly, all of their scores rose dramatically above the claimed average of 20. The average peak score was in the mid-40s. The scores dropped slowly and all but one maintained scores above 30.
Birthday bump versus consistent posts
Several of the undergraduates had birthdays just after signing up for the study, and their Klout scores reached the high 40s and dropped slowly thereafter. This suggests that the increased activity surrounding major life events considerably affects the Klout score.
However, two subjects did record scores in the 40s without the occurrence of any major life event. Instead, these subjects recorded consistent posts and Likes on Facebook pages between themselves and around three friends. Just one or two posts reduced the level of decay or even increased Klout scores, indicating that Klout does value consistency over one-off events. Still, the birthday bump is a source of bias.
Undergraduates on Facebook versus Influencers on Twitter
Klout accounts with only Twitter logins required thousands of followers and high levels of activity to achieve comparable scores to Facebook only accounts with far fewer friends and much less activity. During the tracking period we followed a highly active Australian social media expert with 1800 Twitter followers and hundreds of retweets and mentions. As shown below, with some variations one undergraduate (red) had a higher score but eerily similar decay to the social media expert (blue), and they ended at around the same score of 45.
Similarly, as the graph shows below, the average peak Klout scores of the undergraduates were also remarkably similar to the average of Gruen Transfer panel members. In fact, were it not for Wil Anderson, the undergraduates would trounce the Gruen Transfer group.
Todd Sampson is just below the average of these undergraduates (42), and Russel Howcroft is, apparently, far less influential than all the undergraduates (35).
Nice work if you can get it
Clearly, Klout scores should not be taken at face value. NYU Professor of Business Sinan Aral’s recent Science article argues that online social influence is more about measurable action taken on posts, rather than the number of followers or posts themselves.
We found that Klout favours Facebook over Twitter. Klout may argue that this is because Facebook connections are deeper and more personal than Twitter. However, what it really means is that Klout rewards users who join and connect their Facebook account. This is not a measure of influence, then, it is a measure of willingness to join Klout.
This still leaves the problem of purporting that an undergraduate has as much influence as Todd Sampson or far more than Russel Howcroft. The biggest problem with the Klout score is that it has no idea who people are or what they are saying.
The Gruen members are all are prominent in their field and have significant media exposure. The Klout score apparently does not attempt to include measures of social capital such as employment or media coverage and has no concept of the meaning of posts.
Klout is not really measuring influence, rather, it is measuring activity on social networks.
Brands that want to use Klout should consider the scores only as a rough measure of activity. Premiums should target those with scores of 60. Scores above 10 will capture all Klout users. Scores above 20 will capture very sporadic Internet users. Scores up to around 45 or even 50 will capture users with average levels of visible Internet activity. These are potentially useful guides to likely activity, but they are a long way from a measure of influence.