My Experience with Twitter, Part 3
A few weeks ago, I posted on my experience with Twitter, Part 1. That post was retweeted by Robert Scoble, the traffic came, got a bunch of new followers on Twitter (welcome folks), and a flurry of passionate comments on the post, including 3 comments from Guy Kawasaki. Then, I posted on my experience with Twitter, Part 2. This post is Part 3, here is Part 4, in case you were interested. This post includes a quick analysis on the top-100 folks on twitter.
There is a camp that argues that the top-100 on Twitter, ranked by followers, is a huge homogeneous blob — that is, this is a group that largely looks the same, that primarily follows each other, and engage with one another within the top-100 in a pat-each-other-on-the-back type of relationship.
Furthermore, this camp argues that Twitter follows a long-tail model, such that “the hits” are the ones with the most followers and that the tail consists of the rest of us. If that argument is true, then Twitter doesn’t really democratize the conversation, but that the conversation remains with the elite.
This intrigued me, so I looked at some data. All data was as of 01/22/2009
Methodology and Data Sources
I wanted to learn about the behavior and attributes of the top-100. So, I went to Twitterholic, pulled their data into a snappy spreadsheet, and played with the data.
Discoveries and Hypothesis
There were several things I wanted to learn more about or, even conclude, from the top-100 list. Below are the research questions I had in mind:
- What is the Demographics of the Twitter top-100?
- Does tenure on Twitter likely lead to more followers?
- For the top-100, what does the cumulative distribution and shape of updates-per-day?
- What is the impact of Twitter updates-per-Day on followers?
Let’s address each question in succession.
Top-100 Twitter Demographics
The criticism of the top-100 as a homogeneous group really points to its demographics and, or, its psychographics. But, the best data I could obtain provided me with very basic demographic information, but is descriptive.
The top-100 is mainly a male-dominated group, with women comprising only 12% of the group and men comprising 58% of the top-100. For this quick study, I define “Company” as an entity — corporate or government — whose Twitter account represents that entity. For that category, I judge that 29% of the top-100 are organizations with a Twitter account. I define “Not Sure” as, well, not sure: this account is @darthvader — I know he is a male, but he’s also 1/2 machine and he’s also fictional. I’m not sure how to categorize him/it, so I created the “Not Sure” category.
Other interesting information would be to study the occupation of the top-100, so see if they are homogeneous in that respect (are they all social media journalists or consultants?); other attributes would also be interesting, including geography (are they all in silicon valley?), etc.
Role of Tenure on Twitter
We wish to know whether there is a relationship between Tenure and the number of Followers on Twitter. To do this, I obtained days on Twitter for the top-100 from registration to current day (as of 01/22/2009). Below is a regression between the variables of tenure and followers:
Based on the best-available data I had of the top-100, it appears that Tenure almost has nothing to do with the number of followers.
Caveat: I grant that I am not an expert on the regression (though, my regression on Snoop’s “Ain’t nuthin’ but a G-Thang” is pretty darn creative); I also grant the role of outliers — for example, if I took out the outliers, the relationship of Tenure-to-Followers might actually become stronger, based on the R^2. I considered all of that.
What does Updates-per-Day Look Like for the Top-100?
Earlier, we looked at Guy Kawasaki’s tweet behavior over time. Do the rest of the top-100 tweet as much as he does?
For the top-100, the data above shows that the average updates-per-day is 7.7, with an average spread of 12.4 updates-per-day. The cumulative shape of updates-per-day for the top-100 can be approximated by a normal distribution.
The ones that update the most from the top-100 are below:
- @alohaarleen, 99.2 updates-per-day
- @nytimes, 38.39 updates-per-day
- @chrisbrogan, 37.59 updates-per-day
- @perrybelcher, 34.63 updates-per-day
- @guykawasaki, 33.65 updates-per-day
- @newmediajim, 31.35 updates-per-day
- @breakingnewson, 23.64 updates-per-day
- @scobleizer, 21.73 updates-per-day
- @mashable, 19.13 updates-per-day
- @loic, 17.71 updates-per-day
What is the relationship between Updates-per-Day and Followers?
Similarly as before, I understand the role of outliers and how that can skey the data, but I chose not discard the outliers in this case (mainly out of laziness). If I were actually getting paid to write this post, then I might have injected some data integrity and professionalism; instead, I’m opting for just plain fun and, hopefully, some traffic and a lively conversation.
So,
If I removed the outliers, the shape of the data shows that there might be an inverse relationship between Updates-per-Day and Followers: in other words, the more you update, the higher the likelihood that you will have fewer followers.
Again, with a more professional approach (and if some monetary reward was involved as further incentive), I’d find the inflection point at which updates-per-day begins to degrade the number of Followers.
This inflection point might be interesting to some or, might help some in their communication strategy — “when is overcommunication reached?” — for example, might be a very interesting question to answer for corporations on Twitter.
Conclusion
This was fun and I’m still very impressed with Twitter.
If you’re interested, you can download the twitter top-100 spreadsheet here. Enjoy!
WP Cumulus Flash tag cloud by Roy Tanck and Luke Morton requires Flash Player 9 or better.
If you enjoyed this post, please consider to leave a comment or subscribe to the feed and get future articles delivered to your feed reader.
Comments
http://tinyurl.com/d5bxk7 100 Twittering almost 100 times per day? can’t imagine anyone would follow me
http://tinyurl.com/d5bxk7 More reason to pay attention to your Probability and Statistics course ! !
@ethanbauley you might enjoy this one on twitter: http://is.gd/hiD9 & this one that chris anderson emailed me on: http://is.gd/dckN
Reading @shmula’s analysis of Twitter http://is.gd/hiD9 and remembering my post on twitterverse demographics here: http://is.gd/ajwx
Now this I like: Blogger does statistical analysis to answer the question, How many tweets is too many? http://bit.ly/McOr
I am very interested in what I see here, but one thing that I don’t see is amount of DM replies made to followers.
I spend my days not bothering my whole follower stream with too much personal or unnecessary information. Instead I DM my followers and let them know someone is there and answer questions, comment etc.
I think this strategy is very important in the growth and true “Community” building of what @Foodimentary is to the user.
I make a point to respond to 90% of my replies and every DM I can. This core value is what I believe helps my user growth and retention.
That and of course some darn neat food facts and trivia.
Has anyone factored this in their numbers?
FYI I have made a total of over 3800 DMs in the last 94 days.
I also use @FoodimentaryGuy, a second twitter name, as a way to communicate to my hardcore followers who want to see the funny responses and info I receive
I also would be interested in how retweeting effects the stats for I believe that is another key factor in growth and retention.
Let me know
John-Bryan Hopkins
The Foodimentary Guy
Leave a comment
Additional comments powered by BackType
My theory on twitter is that it enables a cyber-village on the internet