The State Of The Twittersphere: Don’t Look For A Bell Curve
What annoys me about the State Of The Twittersphere post by Erick Schonfeld and the original data offered up by Hubspot is that it doesn’t actually do any meaningful analysis on a bunch of simplistic raw data:
[from The State Of The Twittersphere (HubSpot Edition)]
How many followers do most people really have on Twitter? The average number of both followers and other members people on Twitter are following is about 70, according to the State of the Twittersphere, a new report by Web marketing startup HubSpot. (Full report embedded below). But that average is skewed by elite Twitterers who have hundreds or thousands of followers. The vast majority of people on Twitter use it to keep in touch with a much smaller circle of friends and peers. For those with 50 or fewer followers (three quarters of all users), the average number of followers is 15.6 and the average number of people they are following is 18.4.
A quick read of this paragraph might lead people to target these mean averages for their Twitter experience, which I think is wildly misleading. I have suggested for a longtime that to ‘get’ Twitter you need to follow 100 people at least, for several weeks. This cursory recitation of stats suggests that there are thousands of users out there happily communing with a handful of friends. I don’t buy it. I bet most of those accounts with small use, small links, and small time online represent a fringe of uninvolved people who aren’t getting much value from the service, if they login in at all. The sweet spot is far north of the center of some bell curve, I believe.
The real analysis of meaningful trands will have to wait, but here’s some cross tabs that would be interesting:
- What’s the distribution of perceived value? Does more use translate into higher perception of utility? My bet is yes.
- What’s the distribution of use? Do people with few connections use the service less? My bet is yes.
- Do people gain more followers based on hours online, and numbers of Tweets? I bet yes.
- Where is the magic dropout number? A lot of users abandon services like Twitter, but I bet that once you have a network of size N, the likelihood of dropping out decreases dramatically. What is N?
Here’s a nasty freehand drawing of what I am suggesting:
Utility Graph in Twitter, originally uploaded by Stowe Boyd.
On the left, the vertical access is some formula based on an aggregation or ratio of following, followers, numbers of tweets, relies, direct messages, time spent usign the tool, etc. The horizontal access represents perceived utility. I have drawn the utility curve as being exponential, but it may actually be more of an S curve, with utility tapering off after some psychological maximums have been reached.
Since TechCrunch and Twittergrader are apparently interested in statistics, maybe they could underwrite some serious research here. I’d be glad to help.
