Klout

» see klout.com as it lived, on the wayback machine

~ the goodbye ~

by the time klout would have made sense for me to sign up to, the score had already lost most of the cultural weight it briefly had. klout launched in late 2008 from joe fernandez, who came up with the concept while recovering from jaw surgery and unable to speak for three months. the product gave you a number between 1 and 100 that was supposed to measure your influence on social media. you connected your twitter and facebook accounts. klout watched your follower count, retweets, replies, and reach. you got a score back. higher number meant more influence.

the score was the entire product. there was a dashboard around it and a list of who you supposedly influenced. a perks system existed where brands gave free products to high-score users in exchange for tweets. the dashboard was interesting for the first ten minutes, and the perks were interesting if you were getting them. the score itself was the thing people came back to check, the way you check a credit score, except it updated daily and the rules were opaque and famously gameable.

this was a US tech-marketing product. i am a turkish computer engineer who spent the relevant years writing code in ankara, not auditioning for an influencer-marketing campaign in a US zip code. the score was calibrated against english-language twitter and US-centric facebook activity. for someone outside the english internet it produced numbers that were technically valid and culturally meaningless.

~ the chronicle ~

~ the score ~

the math behind the klout score was the part that got the most criticism. klout claimed to use machine learning over hundreds of signals across multiple platforms. an independent analysis found that the simple log of your twitter follower count explained about 95 percent of the variance in scores. all the extra signals were doing very little. the rest was noise dressed up as sophistication.

this did not stop the score from showing up in places it had no business being. marketing agencies pitched campaigns based on klout cutoffs. there is a documented case of a recruiter telling a job applicant his klout score was too low for the role, and people put the score on their dating profiles. cory doctorow wrote a column about how the score was an early example of letting an algorithm decide who counts. klout was both bad math and a real cultural object for about four years.

~ the perks ~

the actual revenue model was perks. brands paid klout to ship free products to high-score users with the implicit understanding that those users would post about the products on their social accounts. the perks ranged from sample-size cosmetics up to occasional flights and hotel stays. by may 2013 klout reported users had claimed over a million perks across more than four hundred campaigns. the perks system was effectively the first programmatic influencer-marketing platform, four years before the term influencer marketing fully existed.

the perks were also the part that exposed the score's incoherence. high klout scores went disproportionately to people who tweeted a lot about brands, which is exactly the population the score should have been ignoring if you wanted authentic reach. perks rewarded the gameable behavior, and the gameable behavior raised perks-eligibility scores. the loop was visible from the outside.

~ what really happened ~

lithium technologies bought klout on march 27, 2014 for $200 million. lithium was a B2B social-customer-service company and the acquisition logic was that lithium would feed enterprise customers a klout-derived feed of high-influence users to prioritize for support. the consumer score side was no longer the strategic asset. fernandez left lithium in 2015. the klout product slowly stopped getting product investment.

on may 10, 2018 lithium announced klout would shut down on may 25, the same day gdpr came into effect in europe. the timing was almost certainly not a coincidence. klout's data model assumed broad scraping of public social media activity that was much harder to defend under gdpr's data-minimization rules, and the company had already publicly admitted the consumer brand was not a strategic priority. lithium shut the consumer site cleanly on the gdpr enforcement date and kept the underlying influence data for its own enterprise products.

~ what remains ~

the klout score is the cultural artifact that survives. nobody talks about it now without irony, but everybody who lived through 2010 to 2014 has at least one specific memory of someone who took it too seriously. the broader idea that a third party can score your social influence and sell that score back to brands did not die when klout did. it just moved up the value chain into the closed dashboards of platforms like aspireiq and creator.co, where the scores are not exposed to the public anymore but the math is functionally the same.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »