Business Week ran a story this week about Google's patent-pending method for ranking influencers on social networks. Google wouldn't talk about it but Heather Green pieced together the following hypothesis about their criteria:
- How many friends you have
- How many friends your friends have (G1+G2)
- How frequently friends post on each others "sites"
- Click-throughs from recommendations (e.g. if I post a link to a cool video on my wall, how many people click through from their News Feed
Sounds plausible especially when you have the limitless bank of Google behind the effort. I have certainly been waiting for Google to blow Technorati's "authority" ranking out of the water. That much they could do and will do. But the 'magic number' for your overall influence? Not that easy.
Influence is Complex and Context
The problem is that the reported approach doesn't take into consideration the web of context we all float around in. If somebody wants to know about digital influence or Nick Cave - I am definitely their guy. Cricket? Not so much. I have influence (and authority) on some subjects with some people in my network and not on other subjects. How do you map this context so that I as a marketer know to engage (or if you are Google - display advertising on) an influencer and her social graph on one brand or topic versus another?
It reminds me a lot of what 33Across is trying to do. Essentially "mapping" my social graph is highly useful. Weighting all the connections against topic areas is mindboggling. It would take a careful analysis of content to get a sense of what different people are actually authorities on.That pokes a sharp stick at privacy issues (yes, I know it's public information but it's still invasive to inventory it)
Our model for assessing influence is complex and includes between 4-7 criteria some of which are human, (see illustration above) qualitative assesments. We use as much technology as is available to simplify the task but it cannot be fully automated. One small tech innovation that helps is the little slider bar inside the cgm monitoring tool, Radian6 that lets you adjust the weights of different data points to determine influence on a client and topic-specific basis.
Google has a lot of smart minds thinking about this. They will come out with something that is more useful than that Technorati number (warning: that is not a sound measure of influence all by itself). But beware a magic number, it contradicts what we understand about contextual authority.