The Online Influence Players

In the past couple years, a number of players have come to prominence in the Online Influence space. Some are providing free influence rankings and tying into third party applications (like Klout with Hootsuite), whereas others are offering paid services or bundling an influence component into existing solutions (like Lithium’s sCRM).

In this post, I’ll attempt to briefly describe the methodology of a few of the players and to consider the question of who’s doing it right/best.


Klout has been a dominant player on the influencer identification and ranking scene for some time now. Their fundamental ranking consists of three factors which they describe as True Reach, Amplification Probability, and Network Score. True Reach takes a stab at “engaged followers” which is a step ahead of merely counting followers and following. Amplification is derived by second level activity such as mentions, retweets, and likes (since Klout recently added Facebook integration to its predominantly twitter based system). By their own description, the score is weighted towards “clicks, comments, and retweets.”

Klout has won some significant battles in establishing connection with 3rd party software like HootSuite to include their influence ranking within the context of the software. Ultimately, I think this is where tremendous usefulness lies – as with anything in Social Media, you need to bring in in line with existing business processes so that people don’t have to go out of their way to use it.

For more on Klout’s ranking – see their description.


PeerIndex makes an important distinction in their determination of online influence and authority. They recognize that online authority is contextual – i.e. being an authority on one topic has a very low correlation to authority on another topic. So, PeerIndex built their system to determine authority within a given category. This is certainly a step in the right direction.

Their core components on these topics include “authority, activity, and audience.” They base authority on content shared (whether created by the individual or not…) and do a comparison against others sharing on the same topic. This is another important factor as influence is relative. They look at your following for their audience score but attempt to normalize that often deceptive metric by looking for spam, gamed, and inactive accounts and pulling them from the overall audience metric.

Their activity score is based on volume for a topic – too high or too low relative to others on that topic is bad and being the first to mention something is good.

For more on how PeerIndex looks at influence – see their description.


Traackr is a little bit more secretive than the first two in how they calculate influence. Their scoring system centers around three components – Reach, Resonance, and Relevance. Looks like a good start, but it’s hard to decipher what’s going into those scores other than “normalized metrics put through a rules engine” – which would basically describe any influence ranking. This is hardly surprising though as Traackr is one of the “closed door” influence systems with a paywall coming before you see any output or data on their system. It’s certainly a different approach than the freemium models adopted by Klout and PeerIndex.

Click here to see more on how Traackr looks at influence.


TunkRank is one of the more transparent systems for calculating online influence. I’ll let them speak for themselves to some extent – there are three key assumptions they’re making:

  • Influence(X) = Expected number of people who will read a tweet that Xtweets, including all retweets of that tweet. For simplicity, we assume that, if a person reads the same message twice (because of retweets), both readings count.
  • If X is a member of Followers(Y), then there is a 1/||Following(X)||probability that X will read a tweet posted by Y, where Following(X) is the set of people that X follows.
  • If X reads a tweet from Y, there’s a constant probability p that X will retweet it.

These assumptions are then brought together in this equation:


TunkRank compares its methodology to the PageRank Algorithm which is heavily based on Eigenvector Centrality (for more on Centrality, click here) – and so is looking at the relationship dynamics behind influence. This is a key differentiator for TunkRank.

To lay out their approach more simply they claim:

  1. The amount of attention you can give is spread out among all those you follow. The more you follow, the less attention you can give each one.
  2. Your influence depends on the amount of attention your followers can give you.

For more on TunkRank check out this page and this blog.


Though not solely an influence ranking provider, I have to give mention to Lithium in this discussion. Their principal scientist Michael Wu (@mich8elwu on Twitter) has done some of the most impressive and comprehensive work I’ve seen to date on online influence. Lithium builds their influence ranking into the S(ocial) CRM solutions and hosted community solutions they provide.

For one of the most comprehensive discussions of influence available online – check out this series of posts from Michael Wu.

So that will conclude this discussion – thanks for reading all the way to the end. There’s plenty more to say about online influence and I know there are players I missed. What’s your take? Let me know in the comments.

You can leave a response, or trackback from your own site.
  • azeem

    Thanks for the shout out for PeerIndex.

    There are only so many ways of addressing the influence / authority / opinion leadership metric. And a single metric will rarely capture it (how do you compare @kimkardashian to @gcluley? I know which one I would rather trust on technology.

    And that in itself starts to talk to the value of such metrics — best used for decision making. Should I read this tweet or not? Should I offer this person offer A or offer B? A single score — beautiful in its simplicity — is like any single variable, crude in its application.

    We’re trying to find the balance between the elegance and simplicity of simplified scores and the depth and nuance of what you really need to make a decision about something. I’d welcome your thoughts on that. I don’t think we quite have the balance right yet, but we’re getting there.

  • Matt Kammerait

    Great points Azeem. I think that what we’ll find over time is that we need more of a continuum of scores than just a single score. Each with its own contextual value. Unfortunately, you probably also have to trade some mass appeal as people are looking for the silver bullet – the one score to rule them all :)

  • Gary Lee

    Great summary for various products in the market today. To weigh in on Azeem’s points, I agree. Walking around debating whose score is higher is reminiscent of the classic line from Spinal Tap “It Goes to 11″ where a new line of guitar amps was better simply because it had one more number on the volume dial. Influencer scores by themselves do not tell the whole story and can be misinterpreted and misapplied.

    Further, Azeem points out the critical nature of using the topical view of each person to derive influence. Without topical relevancy, I argue that someone cannot be an influencer – something Azeem points out in his argument I think. At mBLAST, we are talking about this in our blog and in our products, and are providing solutions to help marketing professionals better uncover true influence.

    This is an evolving space and an important space. Track us on Twitter (@mBLAST) and on our web site – we’ll have a lot more to say in the weeks ahead as we launch a new product for discovering and tracking influencers for marketing professionals.

    Gary Lee

  • Josh Letourneau

    Outstanding discussion of the “Online Influence” players. As with most “scoring engines”, we’re best to utilize them all. From there, we can see the names and/or profiles that consistently emerge. That’s the rub.

    If you’re on the “list” in the first place, it’s safe to say you’re an “Influencer” of some kind . . . and this is to be valued. I have to mention Duncan Watts here, particular his rebuttal of Gladwell’s “Tipping Point” philosophy in which individuals at the top of the ‘Influence Pyramid’ cascade changes (or trends) through a market (i.e. Hush Puppies).

    However, here is an area where I personally spend more attention . . . and I would like to see actual Vendors concentrating as well:

    What is our sphere of influence? If I’m on the list, can you identify my sphere? Can you identify my network out to my 2nd degree? Can you map it? Surely, the depth and breadth of my 2nd degree (my “sphere of influence”) is more important than the number of comments, mentions, or incoming links, right?

    Further, let’s assume that several individuals all influence within the same small world (i.e. those that often mention and participate in conversations mentioning the hashtag #SNA). Perhaps we amplify each other’s contributions, posts, etc. within the same sphere . . . but are we duplicating efforts? Are we reaching individuals beyond our cluster; beyond our own echo chamber?

    Here is an example: My focus in on the business application of SNA, particularly in regards to Business Performance. I focus more on value creation than purely the science itself, and further, I work in the HR, Talent Management, and Organizational Development sphere. I am a player in regards to #SNA conversation online, however I’m more central to #SNA conversation within the small worlds of #HR conversation.

    So while I may be hypothetically non-influential to overall #SNA conversation (i.e. a 30/100), I may be extremely influential within the sub-cluster #SNA conversation within the broader #HR sphere (i.e. a 95/100). And this is precisely where it gets challenging to provide an “Overall Influence Score.”

  • Martin Hill-Wilson

    Thank you for introducing me to more of the ‘reputation’ slot machines! I do get a kick out of tapping in my online IDs and seeing where I end up on the leaderboard.

    Now I know that the post and responses were serious from serious folk who clearly take the topic seriously. But as I get further into the topic, the less I trust its real world value. That said I’m 100% signed up to Influence and Reputation driving people’s behaviour and creating the cultural patterns we fit into.

    But in all the discussions I witnessed so far, there is a ton of assumption, very little exposed framework upon which all this hangs (not the algorithms but the core observable psychology) and a rather geeky lack of awareness that human behaviour has such important subtleties within its general characteristics that instinct says it will never be reduced to code and remain true to its real world equivalent.

    True to some degree? Or just an ignorance observation?

  • Pierre-Loic Assayag

    Thanks for the mention on this short list of other cool companies in our space! Don’t read too much into the fact that we don’t share much info on our site about our scoring. Our website is just up for an overall and we’ve been too busy to update it (coming soon!)… Now if you want to find out more about our methodology, you can check a couple of posts on our blog:
    - the underlying principles of our scoring system:
    - A discovery driven approach to measuring influence:

    You can also see the presentation we did at TTMM in Toronto on the challenges of measuring influence:

    I personally don’t buy the ‘secret sauce’ pitch around influence measurement and we’re doing our best to be as transparent as possible on what goes in our scoring. Clearly we’re not doing enough yet :)

    Now onto the conversation: I want to comment/reach on 2 aspects of these great comments:
    1- on relevance: we seem to all be in violent agreement that influence is highly contextual. I would even go one step further and say that this context ought to be dynamically defined by the person searching influencer (and not pre-determined) to be accurate.
    2- on Social Graph Analysis: let me throw a rock in the pond on this. SNA is conceptually very appealing but it suffers from 2 major flaws: for one thing, it’s a brand new discipline. The first real research papers on the topic date from the mid-90s, which in internet time sounds like an eternity, but in research time is very very new, especially because until web 2.0 there was no way to test most of its theories as large social network data sets where not available then for mass processing; which leads me to the second issue: data availability. In order to make SNA meaningful, one needs access to web users’ social networks, which for most platforms require the user’s consent (introducing a huge bias in your processing of data) or, if you’re Facebook, means that you accept to limit your analysis to the platform you control, risking the fact that your data are partial or that your users ‘pull a MySpace’ on you.

  • azeem

    I would draw a distinction between various types of service and where they operate in the value chain.

    A great example is in the evaluation of financial risk, where Moody’s, S&P and Fitch had a series of methodologies which they used to systematically measure–at a certain level-the risk of a particular instrument.

    The purpose here was to create a lingua franca that allowed parties to quickly assess a landscape. But beyond the simple S&P ratings should have gone deeper analysis (‘multi-dimensional analysis’ as some might say) that took into much more account the risk profile of the individual investor, as well as their own diligence–for example, doing their own creditors research on a particular bond.

    The lessons from that story are the the value chain can split between service providers who lubricate a very large number of transactions with broadly comparable metrics; and those who do bespoke research for very specific client needs.

    Ultimately they serve different purposes. I fully expect we’ll see companies adopting one of those two positions as we move forward. In other words, are you fundamentally a data business or are you a client business?

    As for social network analysis, it’s been proven at Web scale for a decade now courtesy of PageRank within Google; and Facebook newsfeed personalisation, so it’s pretty robust–and it’s limitations (and benefits) are well documented.

  • Adrian Swinscoe

    Hi Matt,
    Thanks for producing a great overview of some of the major players in this market.

    Forgive me if this sounds a little dense and if I have missed the point but my main concern with all of these measures is ‘What are they really measuring?’ Are they comparing apples with oranges? Is their definition and understanding of what influence is the same?

    If so, what is influence?


  • Olli Parviainen

    Good points on the fuzziness of the metrics. All of those services make a distinction of influencers and influencees. Influence is not a property of an individual but a process between two or more people through coercion, domination, manipulation, clarification, advice and providing prototypes of imitation.

    The twofold distinction Josh made about influence on local spheres is not new: Robert Merton separated locals and cosmopolites (Merton, Robert 1957: Social theory and social structure. Free Press, Glencoe, Illinois ) in matters of influence. The locals have a lot of strong ties within a distinct group while the cosmopolites have ties between these groups.

    From a network perspective this means that cosmopolites can expect a lot of activity and retweets from their actions but locals influence to “followers” is more like raising awareness and discussions. They both have influence over people but different kind. Mind you that this research has been repeated in psychology and sociometry.

    Why this is important? These single scores mean very little unless put into a context. That is why SNA is – to my opinion – paramount for distinguishing Adrians apples over oranges.

  • Influence in Context | Ressive Networks

    [...] is also a fundamental shortcoming of many of the online influence players – they create universal influence scores that are not contextual. They perpetuate this error [...]

  • 纽约留学生

    “Hey – good weblog. Just checking out some blogs, seems a pretty nice platform you might be making use of. I’m currently utilizing WordPress for a couple of my weblogs but I’m not happy with it so far. I am looking to alter a single of them more than to a system equivalent to yours (BlogEngine) like a trial run. Something in particular you would recommend about it?”

    Also welcome you!

  • Stuck in the Middle | Ressive Networks

    [...] interesting phenomenon is emerging in the influence game. As discussed earlier, lots of players – Klout, PeerIndex, TunkRank, Traackr, Lithium, etc. – are making [...]

  • Best of 2010 | Ressive Networks

    [...] 1. The Online Influence Players [...]

  • My Twitterverse: Taming the Twitter Universe! | Tony Vargas – The Virtual Torero

    [...] to at the beginning of this article when I referenced building equity in your personal brand. Well, making use of the planets within your Twitter galaxy to communicate and get your message across can …. This can be important for you if your aim of leveraging Twitter may be to increase visibility of [...]

  • Finding Your Influence Sweet Spot | Ressive Networks

    [...] 2. The Online Influence Players [...]

  • Influencer Whisperers | Ressive Networks

    [...] attention has been given to Influencers (in fact, we’ve been part of the discussion – click here to join in). Many tools and algorithms have been developed to sift through mountains of interaction and [...]

blog comments powered by Disqus