There are a variety of perspectives on how best to analyze influence in Twitter and it is not the goal of this post to present another approach.  Rather, I want to take take account of the existing ones, spend a few minutes presenting a perspective and hopefully get a few subsequent subsequent perspectives on this.

Let me start out by stating the motivation for this post: I am interested in knowing more about how people think about activity in Twitter and the thought process they go through when they see a Tweet that is relevant to them for some reason.  There is an assumption of a direct correlation between perceived influence and consideration.  For example, if I see someone with two followers complaining about my product I probably won’t care as much as if I saw someone who was a well respected industry voice with thousands of followers complaining about me.

I think this is a pretty safe assumption to make but there may be differing perspectives on this.

The Web Ecology Project

At any rate, let’s dive in and first examine some of the work done by the Web Ecology Project, an “interdisciplinary research group” in Boston who do a lot of data mining.

They have come up with a great research article, Analyzing Influence on Twitter (PDF).  The Project define influence in Twitter as “the potential of an action of a user to initiate a further action by another user.”  Using this definition, they then perform comparative analysis of 12 prominent Twitter users for 10 days, who fall into three broad classifications (celebrities, news outlets and social media analysts).  The research paid particular attention to the outcomes associated with original tweets.

Based on their formative definition of influence, the study concludes,

That social media analysts receive minimal reward for the effort they exert in maintaining a conversation with their followers.  For those users that succeed, most news outlets were more successful at having their content pushed to other users.  Celebrities, on the other hand, appear to inspire conversational responses with their followers, yet with more success than the analysts.

So basically celebrities inspire conversations with their followers, news outlets compel people to share information, and  social media analysts are loudmouths who people generally ignore.  Nothing surprising about these findings.

The real value for me is in the initial categorization scheme.  From a user perspective, it probably is interesting to note what type of influencer is tweeting about you because that is likely to inform your consequent action.

Twinfluence – Influence Analysis Light

Moving on, Twinfluence is an online application that ranks users in Twitter according to three categories:

  • Reach – the number of followers a Twitterer has (first-order followers) in addition to all of their followers (second-order followers).
  • Velocity – An average of the number of first and second-order followers attracted per day since the Twitterer established their account.
  • Social Capital – An estimation of how influential (how many followers) a Twitterer’s followers have.

Twinfluence offers some basic indexes based on these variables but it doesn’t seem actionable (to me).  The one takeaway for me here is their observation that there is semantic value in examining the influence of a Twitter user’s audience.  The search engine analogue for me is Google’s observation that a page’s backlinks tell only one side of the story, the authority of the pages linking to the page in question are also important variables to be considered.

Klout – Influence Analysis Heavy

Finally, I took a quick look at Klout, a venture-backed Silicon Valley start-up that “measures influence on topics across the social web to find the people the world listens to.”  Like the Web Ecology project, they define influence as “the ability to drive people to action.”

They provide a score based on 25+ variables that is a “numerical representation of the size and strength of a person’s sphere of influence on Twitter.”

Major categories of variables include:

  • True Reach – a variable that filters spam bots and inactive accounts as well as dormant or non-interested followers.
  • Amplification Ability – The likelihood that a message will generate retweets or spark a conversation.
  • Network Score – A measure of how influential a person’s network is.

I was a little skeptical initially based on the breadth of variables they examine but a quick search for my own profile revealed some pretty good information that seemed fairly usable.  These prompted me to sign-up and take a look around.   I like the basic Klout score as a top-line metric to use however there does seem to be a bit of a learning curve because Klout has created an entirely new language to talk about influence that needs to be learned before it can be used.

My Own Fairly Non-Influential Perspective

I’m a little curious if we even need a new set of variables to analyze influencers.  When I look through my own stream of tweets, the first thing I do is see if anything sticks out as being an extreme response.  In other words, I’m not looking for non-intrusive observations but for exclamations about my product.  From there, I’ll generally think about whether to interact or not – but in general I don’t care that much if the person is influential.  In fact, I rarely look at their followers even as a top-line reference before interacting – I just look at their message. That being said, I don’t believe we have the type of chatter that major brands are going to have so some type of categorization is going to be necessary for them.   I’m just not sure how sophisticated it needs to be.

Curious to get some other perspectives on this. . .