The last days I’ve seen a few discussions about the new Klout algorithm. Among Swedes, it seems like many got their scores adjusted: those that were between 65-85 have seen their scores plunge. Those that were lower have often seen an increase.
I’m not interested in discussing whether Klout is “good” or “bad”. It’s a tool, and just like a hammer it can be put to good or bad use. I’ve seen many people upset with Klout being used for bonuses and salaries, but so far I haven’t seen any evidence of that happening on a large scale - actually, everything so far is “I heard from a friend that …” – and I don’t believe any company is stupid enough to hire someone – or give them bonus – based on a metric they really have no insight in (if you know of someone and can put me in touch with them, I’m ready to change my mind here). Some have reacted to Klout being used for perks, like Bal Harbour Shops’ party during New York Fashion week that required a Klout score of at least 40 to be granted entry, but that’s just what celebrity party planners have been doing all along. Sure, they don’t have numbers, but they do take invited guests clout into account when invitations go out. If anything, I’d argue that using Klout adds a transparency that has been lacking until now.
Reading some of the 1300+ comments on the official blog post about the changed Klout algorithm suggests that far too many take Klout score as a borderline life and death matter. To those people I want to say: you’re not judged by Klout score. Your digital presence is.
Instead, I’d like to discuss when tools like Klout, PeerIndex and Empire Avenue are useful and how to apply them. A hammer is great for when you need to build a tree house, but completely useless when you need to knit a sweater.
And I would like to discuss whether Klouts algorithms are functional and how far you can trust them. Me and Simon Sundén had a great discussion today about True Reach.
Apart from the change in overall Klout score, many have also seen their “True Reach” change. Before this change, it seems that “true reach” was actually based on your followers. Martijn at CloudAve wrote an interesting post about the old True Reach algorithm, basically stating that most of your reach was based on pure follower count. But having ten followers with a thousand followers each has a bigger potential of reach than a hundred followers with twenty followers each. This is reflected in the new True Reach algorithm. It’s not perfect and leaves room for a lot of improvement, but it’s better than before.
Klout in the hands of someone who doesn’t know how to look at the scores is like a power drill in the hands of a four year old: worthless and possibly dangerous. When using Klout as a factor, you have to know what you’re looking at. The total Klout score in itself can be misleading, as can the other numbers.
If you’d just look at numbers, you’d see that Swedish digital superstar Jocke (71 K) is more influential than Rand Fishkin (founder and CEO of SEOMoz, 64 K) and Robyn (the artist, 63 K). You’d also see that despite the fact that Jocke has 9200 followers on twitter (a very high number for a Swedish account by a person who isn’t a mainstream celebrity) where Rand has 42 827 and Robyn 134 925, Jocke has roughly the same true reach as Rand: 19 K for Jocke, 18 K for Rand and 17 for Robyn. In reality, we can easily assume that both Rand and Robyn have a higher reach. So how come Jocke has a higher score?
The key here is to also look at connected networks. Adding a network will always increase your Klout score. Jocke has got 12 networks/services connect to Klout. Rand has four. Robyn has one1.
Let’s compare LinkedIn, Foursquare and Facebook (the other services that Rand has connected) to Jocke:
On Facebook, Jocke has got 2202 friends and 291 subscribers. Rand has 105 friends, and you can’t subscribe to his feed. Checking their walls, you can see that Rand mostly posts checkins from Foursquare and rarely gets any interaction on them on Facebook. Jocke posts links, updates and most of them get reactions from his friends and followers. However, Rand is not inactive on Facebook – his company SEOmoz has a page that gets a lot of interaction. But Rand hasn’t connected that page, and even if he had, you can’t get as much data from Facebook about Pages as you can about Profile. Connecting a page does not influence your Klout score as much as a personal Facebook profile.
On Linkedin, I cannot see how many contacts Rand has since he’s not in my network. However, I can see that Rand has 11 recommendations. Jocke has got 70.
Foursquare gives about the same: Jocke is much more active. Aside from mayorships (Jocke 13, Rand 2), friends (Jocke 1000, Rand 32) and tips (Jocke 19 places done by many people, Rand 5 places done by 0 people) you can look at activity and check ins:
To answer that question, you have to ask “in what area?”. When it comes to social media in Sweden (and basically any topic in Swedish) then I would bet on Jocke for influence. But when it comes to SEO there’s no question in my book that Rand is much more influential.
When you’re using Klout as a tool, you have to first define what you’re looking for. Are you looking for influencers within a certain topic? Remember that Klout only lists the top three for a person, and not very precise either. I don’t think people can give you +K in any topic that isn’t part of your top 3, even if you like me have 11 topics in total. But it’s a start – check the people that Klout lists as influential and then see who they interact with and talk to and you’ll get an idea.
Are you looking for people who can amplify your message? Remember that amplification probability is a relative number to how many followers you have. Lady Gaga has an amplification probability of 12, where Jocke has 75. This has to do with the fact that Lady Gaga is followed by 15 million people, and even if many of her followers retweet her, the relative number will still be low.
And one important thing to remember is that Klout only measures reactions. It doesn’t take in all the people who quietly listen and take part of the messages but never retweet or reply.
Having said that, I think that Klout does a good job of using the data they have access to. From the data Klout has been provided about Rand Fishkins and Jocke Jardenbergs digital precence, it makes the right calculations. What you need to check is which sources it gets it’s data from, and make your own judgment.
I believe that Klout will be a good tool when you want to measure the impact of a campaign. Before, you could count how many newspapers wrote about it, and how many tv spots it got. Today you have tweets, blogs and Facebook updates. Counting them is one way of measuring the impact, adding Klout score or those who blog, update and tweet to the equation will give a more detailed view.
I think that Klout needs to solve some problems. Like the fact that you can’t opt out of a Klout profile. I can opt out of having my website indexed by search engines by using robots.txt, and I should be able to tell Klout that I don’t want them ranking me. They need to find a way to gain the trust of the users, and not only the fanatics that sound borderline suicidal by the drop caused by the algorithm change. They need to become better at different languages. When I was twittering in Swedish about the same topics I’m talking about today in English, Klout believed me to be influential about “kitchen, oatmeal and money”.
I know that Simon Sundén is investigating the Klout score and whether the accuracy of Klout can be trusted or not. He’s a very smart guy with lots of experience of SEO and data mining, so if you’re interested in his results I’d recommend you to follow him.
1 Techically, Robyn hasn’t connected anything to Klout. Klout indexes all Twitter accounts when it finds them, just like Google indexes all pages that it finds. However, you can opt out of Google by using robots.txt – but you cannot opt out of Klout.
2 Checked with Follow cost