ReadWriteWeb reported on the popular twitter clients some time ago. It got us this beautiful, Office 2007 generated chart that neatly summarized everything:
Now, here is another Office 2007 generated chart, showing the Twitter Client usage of “Twitter Power Users” (i.e those who had enough followers to be on the Twitterific Top 100 list). Note that I’ve kicked out Bots and other non-humans (ET & DarthVader included) to make it more accurate. Also, note that this represents “ALL” the tweets of those power users.
Both the charts look extremely similar, don’t they?
Now, since RWW’s dataset was pretty randomly chosen, I can consider their data to represent the “typical” twitter user, while mine represents the “power” users.
Key takeaways
-
“Web”, ie http://twitter.com takes a large amount of the pie. This is because:
- Most people find the web interface adequate
- AND, many of the ‘power users’ were early adopters of twitter, using it even before any of the usable clients came along
- SMS Text Messages are waay more popular among the Power Users than the normal ones.
- Twitterific is more popular among the power users (presumably Apple Addicts™)…
- ….while Twhirl is more popular among the ‘typical’ users.
- Mobile Twitter (http://m.twitter.com) is way more popular among twitter power users than the typical users.
- Note that many services used by the power users, such as Dave Winer’s TwiterGram, Snitter, TwitterFeed (which pulls blog posts into your twitter feed) and Hahlo (Twitter Client for the iPhone) are missing from the typical users’ list.
- Also, many services popular in RWW’s list, such as Twit and Twitterfox, are totally absent here.
Top 20
Here’s the list of Top 20 clients, along with their ranks and number of tweets:
| Rank | Service | Rank | Percent |
| 1 | web |
120316 |
59.03 |
| 2 | txt |
19382 |
9.51 |
| 3 | im |
16360 |
8.03 |
| 4 | twitterrific |
13179 |
6.47 |
| 5 | mobile |
12927 |
6.34 |
| 6 | twhirl |
9259 |
4.54 |
| 7 | twitterfeed |
5087 |
2.50 |
| 8 | TwitterGram |
1123 |
0.55 |
| 9 | Snitter |
1018 |
0.50 |
| 10 | Hahlo |
901 |
0.44 |
| 11 | Seesmic |
834 |
0.41 |
| 12 | PocketTweets |
663 |
0.33 |
| 13 | MahaloFollow |
508 |
0.25 |
| 14 |
369 |
0.18 |
|
| 15 | Twinkle |
360 |
0.18 |
| 16 | iTweet |
179 |
0.09 |
| 17 |
158 |
0.08 |
|
| 18 | Tweetr |
150 |
0.07 |
| 19 | ThinCloud |
142 |
0.07 |
| 20 | MobileTwitter |
130 |
0.06 |
Note how the first one(Web) has more than a hundred thousand tweets, while the last, Mobile Twitter, has only a hundred.
More data from this corpus coming soon! (If you need this corpus of about two hundred thousand tweets from top tweeters, email me and I’ll send it over)
12 responses so far ↓
1 Foomandoonian // May 28, 2008 at 8:29 am
I’m surprised. I try and avoid the web interface whenever possible. Not that it’s bad, just because it’s probably the least convenient.
I’m fascinated to see what other results you come up with!
2 Foomandoonian // May 28, 2008 at 8:33 am
Just occurred to me: I use Twitux, a Linux client, but Twitter always says I posted from Web. Admittedly the big clients all get recorded correctly, but this may account for some inaccuracy.
3 Russell Limprecht // May 28, 2008 at 8:36 am
Yovi: I am a big fan of your work. Louis is a good mentor, and annouced you are on the Techmeme leaderboard. Congratulations. I will now attempt to get you a trackback as you requested.
4 Russell Limprecht // May 28, 2008 at 8:37 am
I look forward to your analysis of Friendfeed.
5 Live Crunch // May 28, 2008 at 8:43 am
I like the statistics here, would you say that are very accurate or just some average of what it is and what it could be?
Thanks for the post, don’t hasitate to visit my technological blog when you get a chance.
6 Yuvi // May 28, 2008 at 8:46 am
@LiveCrunch: These are machine generated, so they are pretty accurate. I can give you the corpus to you if you want to verify….
7 Yuvi // May 28, 2008 at 8:48 am
@Foodmandoonian: That’s certainly one of the cause. The other plausible cause is that a lot of these power users were using twitter way before it got good Client support..
8 Charlie Anzman // May 28, 2008 at 9:51 am
Neat stuff Yuvi. BTW - This piece just made TechMeMe. You may want to do a Yuvi StatBot on yourself
Keep it comin’
Charlie
9 Barry Graubart // May 28, 2008 at 11:00 am
Interesting and very surprising IMO.
I use Twhirl when on my laptop and Twitterberry on the Blackberry and only go to the web UI for two things:
1. To navigate friends & followers;
2. To verify that Twitter is down and it’s not just my client app not working right.
Hey - perhaps power users have drifted back to the web client so that they don’t have to constantly ask themselves “is Twitter down or is it Twhirl/Twitterific?”
Anyway - would be interesting to hear from some power users who use the web client to hear why they like it better than the standalone clients.
10 Martin Jamieson // May 28, 2008 at 5:52 pm
Could it be that the power users (especially those that follow everyone back) use the web interface most of the time because of the API restrictions on connections per hour?
Most of these power users may have their profiles connecting through multiple applications, which all reduce their API quota, and while many of us can function quite happily on Twhirl or other apps, setting updates to every 2-4 mins etc. for the power users, due to the massive number of incoming messages they get, it would make more sense for them to use the web interface, which bypasses the API and lets them refresh every few seconds when they are looking to connect with people?
11 Jackie Danicki » Twitter = the iPod of the web // May 28, 2008 at 7:33 pm
[...] read Tweet ListsTwitter Clients used by Twitter Power Users - The StatBot - Fun stats. Visualizations. Leaderboards.Jerry Yang: “We’re Done”UntitledUntitledCzech President Klaus ready to debate Gore on climate [...]
12 Dean // Jun 4, 2008 at 12:50 am
Interesting I quite enjoy reading through these sort of posts when people take the time to put them together. Its nice to see Hahlo 10th on that list and first of the iPhone clients. What period of time does the top 20 list cover? Is that from ALL tweets by the ‘power users’?
Hahlo is currently averaging around 2500-3000 tweets sent per day (up from about 1000 prior to the launch of Version 3), which I would imagine is quite a bit less that the 0.44% reported for ‘power users’. I’m ok with that, afterall it is aimed at iPhone users and not everyone has an iphone, and no one is ever going to use their phone for ALL their tweeting…are they?
Leave a Comment