On measuring influence in the Blogosphere

December 9th, 2006

Matt Hurst has a great example illustrating why measuring influence as inlinks (what Technorati does) is too simple. Here are two blogs, their inlink rank as computed by Technorati, their average daily visits as computed by Sitemeter, and the trend in visits over the past year.

As Matt pointed out, measuring readership with tools like sitemeter is problematic. As I write this I realize that I read Matt’s post through his feed in Bloglines, so his blogs will not have registered a visit.

Of course, it all depends on what you mean by influence which is mostly a function of why you are interested in it. For example, if your goal is to sell shoes, ads in “Pink” probably have more impact. If you want to push your new book “Taxes are evil” then Malkin’s blog is the way to go. So influence also has to be measured with respect to the community you want to influence.

There are lots of other factors of course. Not all links are the same. Some are from pages that are highly ranked and others not. Links may also have a sentiment associated with them. For some purposes, e.g., placing our ad for our new book on taxes, we may want to ignore those with negative sentiment in computing Malkin’s influence.

Finally, some blogs are mavens and some are connectors. Mavens are trustworthy sources of information whereas connectors are good at passing on information and ideas, but don’t originate it.

We’re developing a model combining all of these factors, and some additional ones and will have some papers out soon.