Map of Porter Novelli Twitter folk on 17th Jan 2008

Map of Porter Novelli people on Twitter 17 jan

Marian Salzman (our Global CMO here at Porter Novelli) has had the inspired idea of getting people in the agency to tweet about the most exciting story this week (probably) — the inauguration of Barack Obama

You can see the results of the experiment on her blog.

I’m all for this, of course, for several reasons:

  1. It gets new people onto Twitter
  2. It helps us create a stronger network among Porter Novelli twitterers
  3. It means I can track who at the agency is on Twitter

Simply asking people doesn’t really work. Marian’s scheme gives us a shared sense of purpose, which is one of the criteria for creating a strong community. It’s a much better plan all round.

But — of course — I want to see whether I can track how that works in practice. What do the before and after photos of a strong community look like?

The map above is the “before” map. You can see that there’s some nice heavy clumping down in the bottom of the chart, but a lot of pendants towards the top of the graph — people who are held into the main network mainly because of @robmctree. If he were to leave the network, you can see that some of them would float away to become like the eight other unconnected nodes at the top left.

That’s why his node shows slightly “hotter” (we rank the nodes according to a colour scale from deep red at the coolest to white at the hottest.) The hotter the node, the higher its “betweenness centrality” (basically a measure of its structural significance – if you remove the nodes with high betweenness centrality, the network tends to fly apart faster than if you remove the others.)

@mediaczar (that’s me) is big and white. The size indicates how many other people in the network follow that user (me.) Normally the big nodes in a network have all the betweenness centrality.

What’s interesting is using this technique to identify those (like @robmctree, @noahbanning, @volapuk, @angie_s and @amytokes) who have relatively low link popularity but high structural significance.

To give an idea of how significant these high betweenness centrality individuals are to the network, here’s are two maps: the original map, and a map with the ten highest scoring individuals removed.
Map of Porter Novelli people on Twitter 17 jan

Porter Novelli twitter network with 10 highest betweenness centrality nodes removed
and after.

Like finding “top bloggers”, finding “top twitterers” is relatively easy. But approaching them is harder: there may be more demands on their time, for one thing. But if we can identify those who are less “popular” but who may still carry lots of weight within the network, we may be able to find softer targets for our comms activities.

On another note: this project would never have happened without @mariansalzman, @tikkers, or @zeenat58 who have been tireless on the internal email channels. This only goes to show that mistaking the channel for the network is an error into which we can too easily fall.


  1. says

    Hey Mat-

    Not sure I can handle the responsibility of holding all these people in the network. :)

    Nice graph. Really interesting analysis of the community. Seems like we should be using this with internal comms clients as well to help them better understand information flow through their networks.

    • says

      Actually, the technique we’re using have been used to great effect in fields from anthropology and sociology to management consulting (“Bob, what would the impact be if we downsize Michael?”) Then there’s a really interesting study of the emails sent between Enron employees by Jeff Heer at Stanford.

      The tools and techniques we use are fairly simple, and seem to have some bearing on real life experience! The difficulty is getting the experience that’s required to interpret the data; which is why I spend so much of my spare time with these things!


Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>