Friday, June 12, 2009

Non-state of the Twittersphere

HubSpot’s State of theTwittersphere report, released Wednesday June 10, produced metrics based on 4.5 million twitter accounts. Arstechnica.com did a great job of summarizing the findings (accurately if I may add), but did not analyze these findings. The State of the Twittersphere report, in my opinion, is plagued by one ill metric which confounds important results. “Active” versus “Inactive” twitter users.

HubSpot defined “Inactive” as a twitter user who has less than 10 followers and less than 10 friends and less than 10 updates (All three criteria must be met). Twitter is a service based on a temporal design. In order words, as the famous saying goes, “What are you doing right now?” By defining an active user as someone who has less than ‘X’ criteria eliminates the very essence of the twitter model, time. An active user, as defined by HubSpot, may be a member 3 or 6 months old who has abandoned his account, but still met the criteria. It may have behooved Hubspot to create a criteria based on time rather than activity, to model the very heart of twitter’s service.

With that said, Hubspot produced ill metrics regarding average tweets (.97 tweets per day), life time tweets (average user tweeted 119.34), and following to follower ratio (.7738 OR about 8/10). According to Hubspot, we have more followers than people we follow (about 8/10). Take a look at any celebrity profile for some anecdotal evidence (Ludacris, Perez Hilton, Ashton Kutcher). What’s interesting are the population metrics stating that 44.5% follow 1 or more people and 47.29% have 1 or more follower. Firstly, they never took out “inactive users” (even according to there own erroneous definition). These could be people who joined the service 2 years ago and have not used it since. A user who tweeted 89 times, meet all of there criteria for active, but hasn’t logged on in nine months. Secondly, because they didn’t classify active & inactive while computing these ‘global’ metrics, you can not calculate a true metric of which users are activity reading tweets (following). They calculated a ratio of about 8/10, but this ratio is inaccurate, because of the non-temporal nature of their “active” user definition. What this means is that even though they say for every 10 followers you follow 8 people. Those 10 followers may be dead users who have not read any of your tweets since they last logged on 11 months ago.

What does this matter? It’s important to understand their definition of “active” is not temporal, which is the model twitter has thrived on. It’s also important to understand that the population metrics may include users who have not logged onto the service in quite some time. Furthermore, their “active” user metrics are plagued by the same ill “active” user definition. This translates into a publishing issue, in that a person may have 1 million followers but only some unknown percent are actually reading those tweets. Several questions remain in that: How do you define active? How do you measure whose reading your tweets? And lastly, How do you measure if users are extracting entertainment value from your tweets? Even though you may have 1,500 followers 20 may only be listening.



Bookmark and Share

1 comment:

gricomet said...

Hi your link is to the Q4 2008 report. Tweeted about your blog and noticed the error. Lots of good points on a faulty definition of active users.