Tuesday, May 31, 2011

Clout, Klout and Social Media Influence: An Interview with Klout CEO Joe Fernandez

Last week, I had the good fortune to interview Azeem Azhar, founder and CEO of PeerIndex, about influence in social media. And today I am excited to continue the exploration of influencers and influence measurement with Joe Fernandez, cofounder and CEO of Klout.

Klout was founded in 2008 with the goal of measuring the influence of those on Twitter. I first became aware of Klout in 2009 when people on Twitter began to compare and track their scores, but Klout really burst onto the scene in 2010. That was when mainstream media took notice of how Fortune 500 brands were using Klout to find and offer special deals to influencers in social media; for example, The Palms Hotel and Casino upgrades guests based on their Klout scores, and Virgin America extended an offer for free airfare to promote a new San Francisco-to-Toronto route. (Disclosure: Thanks to my Klout score and love of Disney, Klout furnished me an invitation to a special preview showing of Disney’s Tangled.)

Today Klout calls itself “the standard for measuring influence online” and this isn’t an idle boast, considering almost 2,000 companies are using the Klout API for things like social CRM and influencer campaigns. Among Klout’s clients are Nike, HP and Disney. (If you’re not familiar with Klout, you can view my public profile here.)

While I respect the success Joe and his team have created, I still have some doubts and concerns about the way influence is calculated and used, so I extended an offer to have Joe share more about Klout and the science of influence.

Joe begins his discussion of influence not with an exploration of science and algorithms but where I believe the dialog should start—philosophy and human emotion: “When you think about it, the idea of measuring influence is kind of crazy. Influence has always been something that we each see through our own lens. In that sense, it's not that different than love or jealousy.”

It is kind of crazy, but the desire to measure and use influence is inevitable today because, as Joe points out, “For the first time ever, influence can scale and there is data to back it up.” While social media has done new things for influence, Joe notes that Klout isn’t the first to consider the science of influence. “Our goal wasn't to invent math here. We definitely spent a lot of time looking at all the work that has been done in the past and then thinking about how to apply it in the context of the social web.”

Klout scores became synonymous with influence and activity on Twitter, but the company continues to expand its focus and inputs. In October 2010 Klout added Facebook as a source of data, which required Klout to consider how to recognize and weigh influence in one channel versus another. “We've had a lot of late nights at Klout HQ arguing about things like whether a retweet is the same as a ‘Like.’ It's almost like a currency exchange but much trickier. Many of the actions that drive engagement and interactions on Twitter would make you obnoxious on Facebook.”

So, which is more important to your Klout score—your Facebook or Twitter activity? That depends, says Joe. “We look at each platform holistically to determine what the signals of influence are. We then perform sophisticated analysis to weight the different platforms appropriately for each person.”

Among the criticisms that Klout and other influence tools have faced is that people who blabber a lot and thus get a proportionally high number of retweets and replies can earn higher scores even though they may actually exert little influence over others. Joe acknowledges this is challenge but is confident true signals of influence are locked in the data. “The data is there to begin closing the loop. We look at things like the ripple effect in your network when you ‘Like’ something on FB and whether your content gets retweeted into networks far beyond your own.”

Still, there must be a difference between posting something that people retweet because it causes them to think versus something many people retweet because they merely like it. The most retweeted thing I posted in the past 30 days was a comment about the erroneous doomsday prediction: “If tomorrow is the end of times, I want to thank my Twitter followers for the great ride. If tomorrow isn't the end of times, see U Monday.” That rippled, but did I actually earn influence?

Joe says that depends on two things--what happens next and whether that tweet aligned with topics I typically share. "We don't judge some content as better than any other—we let people do that for us through their reactions. Your example is part of the reason we also show topical influence on someone's Klout profile. If you are looking for a marketing expert, you need to ensure not only that they have a high Klout Score but they are influential on subjects that make sense, such as marketing or social media as opposed to cooking or something unrelated to your company."

I asked Joe about the objections Klout faced when an experiment demonstrated that a retweeting bot could rapidly earn high Klout scores. Joe acknowledges the challenge but argues that not all bots are created equal. “We have whole science team who is devoted to ensuring our scores and analysis are accurate. Bot detection is part of what they are working on. They then try to analyze how being a bot affects your influence. For instance, @Techcrunch may simply be a bot—a feed of their articles—but to say it is not influential would be unrealistic. We try to look at both sides of every issue when it comes to measurement.”

But is TechCrunch’s Twitter profile a bot?  In my definition, it isn't because @TechCrunch shares original and highly-influential content produced by TechCrunch bloggers. Joe approaches the bot question from another perspective; he sees a difference between a bot, which may automatically feed valuable and influential content, and a spambot that merely posts nonsensical or misleading content.  Notes Joe, "Our science team is continuously working on methods to detect spammers and reduce their Klout. This is a challenging problem that even Twitter has not solved yet, but we love the challenge and take great pride in the accuracy of our score."

Another challenge I offered to Joe was to explain how Josh Bernoff, a best-selling author and in-demand speaker, earns a lower Klout score than I do. No reasonable measurement of influence would make this mistake. Joe admits Klout has room to grow in measuring real-world influence: “Influence is absolutely contextual, and that is why we are looking to add more networks all the time. Right now, Klout measures influence on Twitter and Facebook and is not necessarily a good measure of offline influence. Every day we hope to take in more data and become more accurate.”

Joe tackled the offline-versus-online influence challenge in a humorous and thoughtful post on the Klout blog. In it, he explores how it is Justin Bieber can earn a higher Klout score than President Obama. Joe posted, “Justin Bieber is one of the first worldwide pop stars to figure out how to influence his fan base efficiently through social media and he has done it at a level we have never seen happen before. If Barack Obama and his team wanted to be more influential than Justin Bieber online I am sure they could be. However, given the level of engagement of each of their networks and the actions they drive, that just isn’t the case right now.” In other words, Obama can change the world, but he doesn’t engage enough on Twitter to have this influence demonstrated and measured in social channels. 

As an influence tool should, Klout recognizes that earning influence is difficult. Collecting a lot of followers is easy—you can use a number of tools to auto-follow thousands of people. But it is much more difficult to say something that not only your followers will repeat but their followers will. Notes Joe, “We don't think followers are a good metric for influence. The soccer mom with a hundred friends on Facebook may actually be more influential than the social media professional with thousands of followers on Twitter. It all depends on topical expertise and how engaged and influential your network is.”

Joe knows that Klout’s work will never be done. “Our goal is to never be satisfied and always innovate. Our science team is working on better bot detection, spam detection, improved algorithms, and adding new networks. We are also working on making Klout a more social experience—influence is most relevant in the context of your peers, not in a vacuum.” The idea of an influence measurement tool that is social makes great sense; after all, without socialization there can be no influence. Toward this end, Klout recently rolled out a new version with the goal of allowing you to “own your own Klout.”

I appreciate the time Joe has taken to share his insights about influence and Klout. There’s so much more to be explored as our understanding of influence measurement grows, and I hope to continue to engage Joe on the topic in the future.


Terry said...

Good post and I share much of your skepticism regarding Klout and other metrics. However, we pay attention because we crave comparative measures.When we look at at brands (especially insurers), Klout scores correlate very closely to follower count so what does that tell us? We need to come from the direction of "What so we need from an index" rather than "What can we count".

Augie Ray said...

Thanks, Terry. I agree--when it comes to influence we've definitely started in the "what we can count" zone. I'm looking to both Klout and PeerIndex to do a better job of quantifying topic-specific influencers better in the future. My guess is that we'll find many people with a thousand followers are more influential within niche topics than most with tens of thousands.