Wrote about Finding Influencers and defining Social Media ROI – some thoughts where I reasoned  most influence is measured offline and the best way to measure influence is by looking at “demonstrated influence” – aka – speaking engagements at conferences, author or authoress of books, a columnist or member of the press that has been invited or covers conferences – or an organizer of real events – such as a person who would organize a conference or trade show.

As a side point, Google has, since 2004, has been using topical pagerank to organize search results; it was recognized, using latent semantic indexing (creating a vector of your site – and what it is about – and a vector of the content of a page, and what Google thinks it’s about) that you can have different pagerank for a page of a site when the author was considered authoritative for the keyword phrase vs. not, and there was a few Google Patents out on it – but it was hush, hush, and pretty arcane – I recall mentioning it to Bill Hunt of Global Strategies (who I was working closely with at IBM.com, at the time) and he confirmed it.   But that was then.

What about Social Media Monitoring platforms – how do they recognize authority?  They don’t – because they haven’t defined it.   My idea, was to take the influencer list Radian6 provides and merge it with the social graph of those influencers – the result might be more useful than what they currently are producing – but I ran into two issues when I tried to do this with Radian6:

1. The Social Graph is not exported as a .csv file along with the Influencer list – it should be (in a format that makes sense).
2. The influencer list has several duplicates, in short, a lot of noise.

Tried this with influencers for a popular Razor product and here’s what I ended up with out of Radian6 – of the first 20 results (influencers) 13 were totally off topic to Razors and/or had no demonstrated public appearances where influence could be demonstrated.  4 were of the same person, Jon Homfray, who writes for Badger & Blade and appears to be influential in British Crop Production (don’t know if that’s relevant to Razors or not – but hey, at this point – I’d take an influencer, period).  Two others were identified as possible influencers, Boo Radley and David Wessell – but it wasn’t clear if they could be influential about Razors.

So, at the end of the day, manually looking at results from these tools is necessary and they’re full of noise, so my idea would be to more actively sift through results, determine if they’re relevant to the query, and filter them out if they’re not.

Also, it seems to me that Google provides most of the indexable information needed to connect the dots – and that’s what’s missing – connecting the dots.