Jeremy Porter has a post on Journalistics today about influence scoring and the challenges associated with it.  Jeremy’s post does a nice job of pointing out some of the challenges with trying to use influence scores like Klout, PageRank, etc..  Most notably, when looked at them by themselves, they’re not particularly useful because, unlike a search engine that includes both relevance and influence/trust in its algorithm, there’s no contextual relevance.  So Justin Bieber may have a Klout score of 95, but if I’m selling fly fishing equipment,  the guy with a klout score of 20 who only writes about fly fishing and who is very active in a number of fly fishing community sites is much more important to me.

I don’t think this problem is unique to klout…this is a very difficult problem to solve.  Frankly though, given the changed face of media, I’m not convinced it’s even a good idea to rely on fine-grained scores like this at all.  Knowing that one influencer has a score of 64 while another has a score of 78 might be useful in a world where a relatively small set of traditional outlets have significant reach (and you’re going to be extremely high touch with a small number of outlets), but when you have a completely fragmented landscape, you just don’t need to be this fine-grained.  It’s a bit of a dirty word, but frankly in a world where everyone is an influencer and where links and social mentions drive search performance, the biggest issue is scale – like it or not, you have to build a lot of relationships in order to move the needle for the business and spammy approaches just don’t work.  So the challenge is this – how do I build REAL relationships with LOTS of people without hiring an army of people to do it?  When you rely on these fine-grained scores, inevitably you get caught in the discussion of  “is this person really more influential than this person in my niche.”  It’s a total time suck and it really shouldn’t impact how you engage.

Given that you need to engage with a lot of people in order to have an impact, I think you’re better off thinking in terms of broad groupings – i.e., a person’s level of influence is either high, medium, or low.  Then you can focus your efforts on the thing that really matters – developing the processes and tools that will allow you to engage with more people (in a real, relationship-oriented manner).  Specifically, you need to reduce the time required to: 1) find out when influencers are talking about the topics you care about (so you can engage), 2) keep track of the conversations you’re having with influencers (so your conversations are more meaningful and relevant), and 3) engage with more people in less time without sacrificing personalization and relevance.

So, given this, you’re still left with the challenge of developing a methodology for classifying people into the “high/medium/low” influence categories as a starting point.   I think the details for this are probably best covered in another post, but at a high-level I think there are three things you look at:

  • Are they relevant?  (using tools like listorious, alltop, google searches, monitoring, etc)
  • What percentile do they fall into for some of the key engagement and reach metrics? (e.g., average comments, uniques, retweets)
  • Who’s in their network (i.e., do they have relationships with some of the known influencers in the space)?

All of this info is available, the key is developing a way to quickly aggregate it and leverage it to classify people.  I’ll cover this in a follow-up post.

Paul May

Paul May is the CEO and co-founder of BuzzStream. Previously, Paul was the first employee at Support.com, helping grow the company through its IPO. Paul also helped build products and grow revenue at BMC, Tonic, Alterpoint, Wavebender, and Pluck. You can follow Paul on Twitter or Google Plus.

More Posts - Website