Analysis of Twitter Client Statistics

Mark — Thu, 30 Oct 2008 18:15:45 +0000

Tracker has been running since August 6th (almost 3 months) so it’s time that I look at what it’s reporting to see how useful it can be. If you’re not familiar with the methodology that’s employed by Tracker, read the description before reading the analysis. Here are some stats related to traffic and tweet analyzing:

You can see that Tracker has sampled almost 2.5 million tweets, an astounding number. What’s even more amazing is that it’s found that almost 800 unique services have posted to Twitter at one time or another. That’s an impressive figure even if you think you know there are a lot of Twitter services out there.

A previous experiment was conducted by ReadWriteWeb back in April that shows results that vary greatly from my results. Unlike RWW’s analysis, which was for just 37,000 tweets gathered over 24 hours, Tracker’s pool of tweets and timespan is much, much greater than RWW’s. I’m really hoping that this leads to increased accuracy in results.

All figures for the following analysis were gathered from the screen that shows all services for the life of the project. That link can be found here.

Top Services

Web traffic accounts for almost 44% of the updates Tracker has seen. However, this figure is bloated as any service that isn’t recognized gets lumped in as Web. ReadWriteWeb ran an article quoting someone in the know as saying that the API receives as much as five times the amount of traffic the web UI does. For the most part, I take the web’s share with a grain of salt as Tracker was written to track the statistics for third-party clients.

The next three most popular services are TwitterFeed, Twitterrific, and txt. These services have 17.0%, 10.6%, and 10.5% shares on the web-excluded sample, respectively. Rounding out the top 5 clients are Twhirl and TwitterFox with 9.5% and 8.2% shares. The top five clients account for 56% of all known third-party tweets that go to Twitter.

Of the top 10 clients, two are AIR apps (Twhirl and TweetDeck), one is Windows only (Tween, a Japanese service), two can be found on the iPhone (Twinkle and Twitterrific), one on the Mac (Twitterrific), one for Firefox (TwitterFox), and two that are mobile-based (mobile web and txt).

Twitterfeed Reigns Supreme?

It’s interesting to note that the top service requires no consistent manual intervention by the user. Twitterfeed simply tweets your blog posts (from your RSS feed) so you don’t have to. In the words of KDFrawg (amongst others), “Twitter is for announcements”. Auto-fed posts have a lower chance of being followed up on than a physically published post does. Along the same lines, #17 on the list is Twitter Tools, which accomplishes basically the same thing. Way down toward the bottom is mu :(

So what does this mean? Depending on your friends, you can expect to see almost two out of every ten tweets coming from Twitterfeed, which means that automatically no more than 80% of tweets you see are coming from a live human.

Twitter in the Mobile Space

The two most prevalent sources of mobile tweets are txt and mobile web. Together, they account for 9% of all tweets. That’s not even counting how many of the web hits could have been accessed from a mobile browser or the iPhone/BlackBerry only applications. Of the top 25 services, three can be found on the iPhone (Twitterific, Twinkle, TwitterFon) and one on the BlackBerry (TwitterBerry). Just behind that are applications like BrightKite, Hahlo, Loopt, and other primarily mobile applications.

Suffice it to say that Twitter’s presence in the mobile space dominates that of any other major microblogging platform. To top it off, Twitter doesn’t even have an official mobile application like other platforms do (unless you count the mobile interface). I strongly believe that one of the key reasons Twitter is so successful is its mobility. Look at the facts and figures and try to prove otherwise.

Availability Dictates Service Use

When Twitter’s API was restricted to 20 calls per hour, I noticed a major shift in the user base back to the web UI. Once the cap was removed and put at 100 per hour, people jumped back to their favorite third-party client. As of yet, there is no “killer app” for Twitter – at least none that I’ve seen. Each application has its own strengths and weaknesses. Looking back to the first major section of this article about the top apps, you might have noticed that there are no Linux-specific applications on the list. You’ll also see that a good number of applications are being developed on the Adobe AIR platform.

Why air AIR clients so common? Because it’s cross-platform availability on the cheap. Why build a native app that only works on one OS when you can build the same app to work for three? AIR has its advantages and disadvantages, don’t get me wrong, but it’s not going to be going away anytime soon. It’s not because programmers are lazy – it’s because they’re trying to reach out to as many people as they can with the greatest ease.

Tracker’s Future

Once every few months I’ll look at Tracker and see where the ship is headed. I haven’t done any major analysis until now because I’ve been waiting for data to draw conclusions on. It’s hard to look at 2.5 million tweets and tell me I’m wrong, which is exactly the reason I wrote Tracker in the first place: to be your go-to source for client statistics in the world of Twitter.

If you’d like to share a link to this article, please use the shortened URL http://blincc.com/x

mark.bockenstedt.net » Statistics

Analysis of Twitter Client Statistics

Related Posts