Understanding Klout

by Tom Webster on October 19, 2010

Readers of this blog know that I am “ambivalent” about online influence calculators. Setting aside questions about how they are calculated, I think there are even greater questions about why they are calculated. I ask a lot of questions about these things not because I am cynical, but because I am a relentless questioner, as my friends will attest. Questions make things better.

So far, my questions about these services have yet to be satisfactorily answered, but I think a little Darwin will sort all of this out soon enough, with the indefensible measures disappearing and the more promising measures gaining some traction and improving in the process. I had a short but lively exchange with Jason Keath and Jason Falls about this exact subject on Twitter last night. Jason Keath has seen some things he likes about Klout, and he tells me that they are putting a lot of smart thinking behind their bid to become the measure of online influence.

What I like about Klout is their focus on topics – there’s no such thing, really, as a generic “influence” – one can only be influential about something to someone. Still, there is that reductive Klout score, which I believe undermines their credibility with serious marketers as much as it enhances it with the badge-happy classes.

In the well-documented spirit of this blog, however, I am not a doubter – I am a questioner. I think my biggest question is this: is it even possible to measure online influence, divorced from offline influence? If a Klout score is truly focused solely upon online behavior, then Oprah’s Klout score should be far less than 65, since she has only 134 Tweets and follows just 19 people. Clearly her offline influence, not her online behavior, is solely responsible for her higher Klout score. Yes, she has well over four million followers, but certainly not by dint of her Twitter ability!

If offline influence plays a role here, then surely someone so influential as Malcolm Gladwell should pull better than a Klout score of 25? Jason Keath correctly noted that Gladwell doesn’t really have “online” influence, and from a strictly are-you-good-at-Twitter perspective, that is certainly true. Yet, we were all talking about his recent “the revolution will not be Tweeted” piece online just over a week ago, and I would argue that his thinking is extraordinarily and demonstrably influential online. If Klout is working towards capturing that kind of influence – including citations, searches, trackbacks, etc – then they are building something very impressive indeed. SkyNet should be nervous. Jason Keath is right, however – Malcolm Gladwell is not good at Twitter. But is Oprah?

Chris Brogan is good at Twitter, and his Klout score reflects this. In fact, at an 85, he’s 20 points higher than Oprah, despite a significantly smaller platform. Seth Godin is bad at Twitter, and his two, non-interactive accounts (SethGodin and ThisIsSethsBlog) score 33 and 47, respectively. I would argue that the set of individuals more influential to online thinking than Seth Godin would not require ten fingers to count, but again – he is bad at Twitter.

Maybe my issues here are more to do with nomenclature – Klout refers to the Klout score as a “measurement of your overall online influence” and their product as “the standard for Influence.” They may very well be the closest to cracking that nut, but there are a lot of offline confounding variables. The Twitter universe is just too small to serve as the basis for those conclusions, and the whole process strikes me as inductive (rather than deductive) reasoning. After all, if Seth Godin and Malcolm Gladwell have chosen not to interact on Twitter, then perhaps it is Twitter, and not those gentlemen, that is unimportant.

I am not sure what to think about my Klout score. Today, my score is a 53, which is twice Gladwell’s score and six points higher than Godin’s. I’m just 12 away from Oprah. Klout does a very good job characterizing the nature of my behavior (I’m a “Specialist,”) and the topics for which I hold some measure of influence (research, social media, marketing). I think these are bang on, and indicative of the smart things about Klout.

However, though I have a moderately healthy self-regard, I am under no illusions that I slot in somewhere between Godin and Oprah in online influence. If the ThisIsSethsBlog Twitter account tweets that Godin has a new book for sale, he sells a jillion books. If I do, I get a nice call from my mother. If we are interested in the realm of influence beyond the easy, immediate measures (like the confounding variable of the retweet) then the amount of data mining required presents what may currently be an intractable problem.

Another question I have about my Klout score is the nature of its volatility. When I wrote about this topic a few weeks ago, my Klout score was 34. Today it is higher by nearly 20 points. I have been on Twitter since February 6, 2007, so I’ve left a significant, longitudinal trail of data on Twitter. If my influence could really change so quickly, then this suggests two things: one, that the system is probably easily game-able, since I certainly didn’t do anything to consciously goose my score, and two: that “influence” as measured by Klout is an exceedingly ephemeral concept. What is the half-life of influence?

Ultimately, my question is this – are online influence measures such as Klout art or science? The undisciplined answer is that it is both, but that’s like being a little pregnant. One of my favorite definitions of science is this: if you can formulate a hypothesis and then test it, it is science. If you cannot test it, it is faith. How does one test an influence score? How do you know, in other words, if you got it right?

Having said all of that, I remain, as ever, in search of information, not evidence. I am not drawing a conclusion about Klout, because I do not have all of the information. Consider this my plea to be informed. If, as some have suggested, Klout scores are being used by HR departments to vet job candidates, then I think it behooves us all to understand it a little better. There is certainly precedent for measures such as Klout – witness the famous “Q Score” for consumer appeal – so cracking this nut certainly has utility for marketers. Consider this a genuine, open invitation to enlighten me in the comments, which are of course always welcome.

Be Sociable, Share!

http://codygibbs.com Cody Gibbs

I think if we approach these tools with the right perspective, they can be extremely useful. If there’s one thing that Klout does well, it’s separate those who have twitter-botted their way to 10k followers yet have little interaction to show for it, with those who actually engage with their audience in meaningful ways. I also think their behavioral mapping is really helpful in identifying the type of content that a user posts. If they are simply a curator and RT everyone else, Klout will tell you that.

Marketers and HR reps will always look at numbers. No matter how much we try to steer ourselves away, we are driven and influenced by numbers. Better that they be influenced by a somewhat intelligent set of data than just looking at follower count.
Tom Webster

That’s a great point about Klout’s ability to separate the bots from the men (and women), Cody – certainly valuable in its own right. And I know people who use Klout successfully in targeted marketing efforts. Again, I’m on a quest to understand. Thanks for your comment!
http://paulgailey.com Paul Gailey

Whilst the score is of interest I confess I would be more interested in being able to interrogate/query/search by the influence matrix of the 16 classifications they have created. ie Find a high/low scoring “thought leader” or “conversationalist”. I suspect Klout has lots up its sleeve.

One observation I note, is that despite assigning TechCrunch a high score of 97, Klout cannot display it’s page!
http://leighhimel.blogpost.com Leigh

My friend Mark Ury and I had this conversation (on Twitter of course) yesterday when we saw those posts going around. His point (which i agree with) is that Klout doesn’t measure influence as it does engagement. Mark’s quote “tweet often and reply and it goes up. More noise, less signal=good.”
http://www.klout.com Joe Fernandez

Hey Tom,
I am one of the cofounders and the ceo here at Klout. This a great conversation and I appreciate the questions you are asking here.

A couple quick points:
Online vs Offline Influence – This is something we struggle with a lot internally. Conceptually it doesn’t totally make sense to say someone like Warren Buffet who isn’t on Twitter isn’t influential. We are working on some solutions around this where we could say something like “this person is not active in social media but you should know…”. In terms of people like Oprah having a high score but not creating much content, this is also a tough one. If you look at http://search.twitter.com/search?q=%40oprah you can see that she gets tons of incoming traffic but the fact that she doesn’t engage keeps her score lower than one might expect.

Score volatility – One of our biggest challenges has always been keeping everyone’s scores fresh. In the last 3 weeks your data has been processed 3 times while in the previous 3 months you were only processed 1 time. The fact that we are analyzing more data about you gives us more accurate picture of your influence. We are very close to solving this problem and having all scores updated every 24 hours.

Is this bullshit – The question of art vs science is a great one. Influence is such a soft word. We are focused on creating a standard here. I want to be able to say to a brand that if a person of Klout score of 34 mentions your brand you can expect X clicks, Y likes, Z retweets and that is different then a score of 57 and here is the data to back it up. We are not there yet but we do an amazing amount of work to make sure our scores are highly correlated to actions like link clicks and other iterations.

Again, thank you for the fair and very relevant post!

Joe
Tom Webster

I’ll have a think about all of this today, but before I do – I am super glad you posted here, Joe. I was hoping you would.
Tom Webster

Leigh – I think a word like “engagement” may in fact be a better term. Either are better than “trust,” as I’ve seen on things like TweetLevel. Best way to ask me if I trust you is to go ahead and ask me. Not sure how you derive motive from clickstream behavior. In any case, more noise, less signal=good indeed.
http://megfowler.com Meg

While I love a good algorithm, Klout is one I can’t really make sense of — beyond, as you say, making sense of why we need a Klout anyway.

Apart from my horror of ‘C’ words becoming ‘K’ words, my main concern is that the way we communicate as humans isn’t something that fits into an algorithm. Yes, we have patterns, and yes, some things are predictable, but it’s hard to quantify how people react to, and spread what we say.

As a recent example, Penelope Trunk wrote a post on TechCrunch that irritated the hell out of 99% of the women I know — so it got RTed all around the blogosphere, but not because anyone was being influenced… they were just irritated. I suppose that’s a form of influence, but how would the algorithm process the different between negative / sarcastic mentions, and sincere ones speaking to real impact?

There are a million questions, but context is the big bugaboo for me.
Tom Webster

I guess if TechCrunch weren’t influential in some way, then no one would have cared about that Penelope Trunk post. Obviously they are, and lots of women (and men, I might add) reacted to it. In that sense, “influence” is not a positive, but a physics term – can the influencer set bodies in motion one way or the other? I do think that is easier to get at (and by “easier” I mean relatively, as I think it’s plenty hard) than trying to calculate Trust or anything that speculates on why someone might respond or react to things online.

Welcome, Meg! Nice to meet you at IMS in Boston!
http://tmiesen.com Tom Miesen

I don’t take Klout all that seriously. It’s a cute vanity item for those who want to be evaluated, and for all the reasons you listed above, I’m not sure it accurately measures influence.

I am concerned, however, about how recruiters and employers use it. A Klout score is a quick shortcut to actually digging around someone’s online presence and checking out their content. I fear that employers may give too much credit to a Klout score. It’s already hard enough to get a job for a fresh-out-of-college individual, and once you add in Klout score as another potential barrier/filter, it gets even more difficult.
Tom Webster

I am extremely concerned about the HR angle as well, Tom – but again, though I am generally down on this particular category of metric, let’s not forget that Google does the same thing – weighs content by its merits and by how many people “vote” or assign it authority. I think with people it’s an exponentially harder problem – potentially an intractable one – but I am reluctant to dismiss it out of hand. It may not measure influence. It may, someday, measure something. That thing may be valuable. Again, I’m not sure. But I AM sure glad you posted.
http://gearboxmagazine.com Brian Driggs

I guess I just don’t see the value in metrics like Klout.

If they want to be able to tell BRANDS, “If a person of Klout score of 34 mentions your brand you can expect X clicks, Y likes, Z retweets and that is different then a score of 57 and here is the data to back it up.” What value is there for PEOPLE signing up to see their scores? Seems to me like a novel game (popularity contest) to see who can be the biggest targets for archaic push marketing tactics.

Influential people are influential because they DO things. To some extent, Klout is tracking this, but for all the brilliance that is the greater digital community’s ability to come together as one and affect change worldwide, there sure seems to be a lot of effort being put into finding ways to sort us all out into tidy categories.

Who is most influential? Who cares?
http://martijnlinssen.com Martijn Linssen

Fine post Tom, good points. My experience with Klout is that it does a really good job at measuring the weak ties and the little people as well as the strong and big ones – regarding Oprah, just bare in mind they measure Facebook as well as Twitter

However, I’ve written two posts on Klout (http://www.martijnlinssen.com/2010/06/why-i-have-doubt-about-klout.html and http://www.martijnlinssen.com/2010/06/get-your-act-together-klout.html) and they have to do with reliability: since the release of Klout 2.0 (http://klout.com/blog/2010/05/a-revolution-in-klout/) the score is not as trustworthy at all as it used to be

Lots of this had to do with the World Championship and Twitter being really busy, and everything of course with the fact that Klout couldn’t handle that – IMO an app should be able to distinguish between successful and failed results, and not blame other apps for failure and still process the results themselves – without warning even

Now, looking at your score going up from 33 to 53 in two weeks, here’s the biggest cause for that: amplification. From an average (and steady) 40 mention counts since April 2009 you’re suddenly on 250, from 110 last week. Retweeters? Up threefold

That either means you’ve become incredibly interesting all of a sudden, or that you’re tweeting a lot more. As you can see, your relative follower mention% is 3, from 1, where your relative follower RT percentage is a steady 5%

Looking at your tweeting habit (http://tweetstats.com/graphs/webby2001), you’re doing a nicely steady +/- 500 tweets a month, which doesn’t explain your increased popularity. When I check your @replies (one of my private tools), you got 387 over the last 8 days, including RT’s, and that’s just amazing for someone who does a mere 500 tweets a month – then again you chat with some big guys

For comparison’s sake, check out Edelman’s Tweetlevel: they’re doing a fine job as well AND are very steady: they give you a 65 http://tweetlevel.edelman.com/user/webby2001

Measuring influence is science, and Klout as well as Edelman are very good at it, and getting better
Tom Webster

I don’t think anyone would agree that I’ve become incredibly interesting all of a sudden.
Tom Webster

And on the recency effects, Martijn (and everyone else), I wonder if Klout scores decline when conversations with “the big guys” as you put it also decline? There are loads of circumstances under which conversational flow might wax and wane, but my real-life “influence” with the people I chat with on Twitter shouldn’t fluctuate accordingly – as the baseball scouts say, once you demonstrate a skill once, you own it, right? Joe?
http://www.klout.com Joe Fernandez

We look at influence from multiple perspectives. Your overall influence and topic influence will decay over time based on your activity and who you are interacting with. We also measure how much you influence each relationship in your social graph. Those algorithms are less sensitive to time. You don’t necessarily need to communicate with someone every day to have influence over them.
http://www.PromoWithPurposeToday.com Heidi Thorne

I don’t put much faith in Klout either. Also find it amusing that some of the top celebs in both offline and online worlds have miserable Klout scores. And the ones with minuscule followings of a couple hundred and follow only 10 have scores of 41? Really?

Okay, I’ll grant that Klout is measuring based on the influence you have with your particular following, not your overall influence in the community or world.

The only clout (note spelling) that matters is your ability to get people to act… not just how your interact.

Thanks for expressing what I think a lot of us are thinking.
http://martijnlinssen.com Martijn Linssen

Tom, others, please disregard my remark about Oprah: as it turns out Klout didn’t start measuring Facebook influence until last week (http://klout.com/blog/2010/10/do-you-have-facebook-klout/)

By the way, given her 2 million true reach, 0 amplification and 0 network count, I think her 65 score is pretty fair

On a side note, looking at Heidi’s last score (June 20 2010) of 26 (!!!) it seems she was one of the many victims of Klout’s erratic behaviour around that time
Tweetlevel gives her a rightful 63 (http://tweetlevel.edelman.com/user/heidithorne)

Joe, while you’re here, isn’t it time to make up for this? As you can see, this will just keep haunting Klout unless you deal with it
http://www.PromoWithPurposeToday.com Heidi Thorne

Martijn, you are so sweet! Thanks so much for checking out my Tweetlevel score.

You also bring up a good point about different systems of measurement. On one you’re a social media god, on another, well, you might not even rank. Plus, some of these change day to day. It’s kind of like the stock market.

Another issue is that we all need to agree on which measurement system will be the de facto standard. Each industry or community will decide what they consider to be a fair measurement of influence. And that may go way beyond the numbers. It might not even be in numbers. But the only numbers for business that matter are those that follow this symbol: $.

Love this lively discussion! Worthy topic because it can affect, as Tom Miesen points out, employment or other business decisions.
http://argylesocial.com Eric Boggs

Whether or not the Klout number is accurate/precise/meaningful isn’t nearly as important as people attributing value to the number, however misguided the “attribution” or “value” might be.

Smart people like Tom questioning its validity serves as a strong signal of its relevance. And the large distribution amongst integration partners is another powerful indicator of its potential importance.

These things – and the fact that they’re obviously incredibly thoughtful, nuanced in their approach to addressing an incredibly complex problem – lead me to believe that Klout is here to stay.

Doubt that it will become “PageRank for influence”…but it will become something that someone wants to acquire.

Well done, Joe, for addressing everyone’s comments.

Eric
http://www.oneforty.com Janet Aronica

I don’t know what Klout is looking for. I know since I’ve started measuring my Klout score (since I got the @oneforty job and started drinking the Twitter app Koolaid pretty hard) I’ve gotten about 700 more Twitter followers (@pistachio RTs me a lot and she has a lot of followers, that is why) …and my score hasn’t increased at all. Arguably my influence is greater because I have more followers. Now, I’m also Tweeting a lot more at this job because it’s expected of me. So I’m more engaged, definitely. So if it’s measuring engagement or influence I’m not sure but either way do I need to be getting thousands of more followers in order to move the needle?

I’d be weary of hiring someone based on Klout scores. I’m sure that the Klout score is just one small part of a very large puzzle but I think someone’s capacity to grow, learn, adapt to this constantly changing environment…someone’s agility is more important than some a number. Perhaps I’m just a qualitative romantic like that but that’s my take.

Tweetlevel is great: http://oneforty.com/item/tweetlevel
Also see Twitalyzer: http://oneforty.com/item/twitalyzer-2-0

Janet Aronica
Community Manager
oneforty.com
http://www.harechevy.com Indiana Chevy Dealers

Tom thanks for posting what I have been having discussions about recently. Im ok with Klout telling people there habits and with people using the metrics to determine loads of different things. Im not ok with them assigning a number to everyone & them becoming the online police. As many of the commenters have stated measuring people is nearly impossible. Maybe the fact that a post I make from my work account gets RT’d by someone with a high Klout score wont make as a big a difference as someone with a lower Klout score but a following that is more in tune to what my efforts are online based mainly on geography. Used in certain circumstances and in certain ways Klout has some value, just no where near the value that the online pack mentality is giving it.

Chris Theisen
Director of Digital Communications
Hare Chevrolet
http://www.PromoWithPurposeToday.com Heidi Thorne

I think one of the things that troubles people about Klout score is that they feel they have little control over their score. As Janet points out, one wonders what it’s going to take to move the needle.

With Klout measuring 35 variables, where do you start? How much do you need to change those variables to make a difference in the overall score? And since some of the score is based on the actions of your following, it’s really out of your hands. Even worse is that it’s also based on the influence score of those who interact with you.

Chris makes a great point about the geographical nature of scores. While we have a global audience on Twitter, for many businesses, the relevance is in the locality.
Tom Webster

Heidi: I am not so sure that there will be one standard. After all, we do have three credit scores for a reason, and Klout is certainly analogous to a social credit score. Apparently, Lunch is on Brogan.
Tom Webster

Eric – I suspect you are right, though the Klout of tomorrow may or may not purport to be what the Klout of today is. I get where Joe was going with his clicks-per-Klout example, which is at least testable. And thanks for stopping by – we have to have that coffee soon and catch up on Argyle Social!
Tom Webster

Janet – today, I would think that rejecting someone based upon Klout score is likely lawsuit material

I think I am even less sure about Tweetlevel, in that “trust” is a component of their score. I do not believe trust is derivable from Twitter behavior.
Tom Webster

Joe, are there possible negative effects of connecting Klout to Facebook? What if, for example, you are big on the Twitter but keep Facebook for close friends and relatives? Would adding them into the equation “dilute” your Klout?
http://www.klout.com Joe Fernandez

Hey Tom,

No, connecting your FB account will not lower your Klout score. We weight the algorithm based which network you are most active/effective and use the remaining network as additive features.
http://www.squarejawmedia.com BrianMcDonald

I agree with you Tom in that Klout is just measuring Twitter chatter (and now Facebook as I just connected that account) and not all online activity. You could be a great blogger and commenter as well LinkedIn persona and not have that activity influence your ranking (yet!). I also agree that it does a good job characterizing your personal based on the conversations and information it is analyzing.
http://www.chris-moody.com Chris Moody

Tom,

Great post and cool to see Klout jump in the action here. This reminds me of our conversation at BlogWorld as well with Jim Tobin, Kipp, Jeff, DJ Waldow, Jason Keath and a few other folks…

I think that there have been significant improvements in Klout, but everyone needs to have their own definition of what it is. Personally, I like Klout most for how it analyzes my trends and not just the score that it assigns. Mine has gone from 28 to 44 in six months, which hopefully is somewhat indicative of me having a job more heavily involved with social media and networking…

Regardless, great points and great discussion.

Props to Joe for jumping in. I dig the honesty and authenticity.

CM
http://www.greeblemonkey.com Aimee Giese | Greeblemonkey

I am totally with you on being influential about *something* rather than just generally influential. Numbers matter, but it’s what numbers and how. Many of these numbers, IMO, should be just be used to compare apples to apples, and not the sphere as a whole. And *certainly* – as you suggested – the people looking at the numbers should understand what they are seeing.
http://www.cecilykellogg.com/ Cecily

This is a fascinating piece about Klout and I’m grateful that Aimee shared it so I found your site.

I have a pretty good Klout score — 56 — considering that I don’t do much other than hang around twitter swearing a lot. In addition, it calls me a “thought leader” which, again, is kinda hilarious. Because most of my thoughts involve swearing about things (last night it was politics, thanks to the election).

Ironically, however, I do think the score is actually a good snapshot of MY particular influence on MY particular corner of the web — mom blogging. But outside of that community, no one has any idea who the hell I am. So that number is not particularly reflective of my overall influence.

Interesting stuff. Thanks for this.
http://www.facebook.com/roberto.cumaraswamy Roberto Cumaraswamy

Hi there, simple question about KLOUT (as a purported influence tracking tool) – now that they allow you to link your Facebook profile and use the Open Graph tool requiring permission to access your posts, are they tracking your Facebook connections and interactions (within a resticted profile) as well as Twitter?What about the recent inclusion of Linked-In as well?