I looked at all tweets sent in September that mentioned a cable news show, and ranked the clients used to send them by network. For each network, here are the top three clients…
Click to enlarge
| Network | 1st | 2nd | 3rd |
| Al Jazeera | Web (25%) | iPhone (21%) | Website (17%) |
| CNN | iPhone (29%) | Web (25%) | Android (19%) |
| Fox News | Web (30%) | iPhone (27%) | Android (17%) |
| HLN | iPhone (32%) | Android (24%) | Web (23%) |
| MSNBC | Web (32%) | iPhone (24%) | Android (15%) |
| All Networks | Web (28%) | iPhone (27%) | Android (17%) |
Observations:
- HLN’s mentioners were the most “mobile”. There’s not a lot of them, but they’re on their sofas, phone in hand.
- MSNBC’s were the least. Either they’re all watching MSNBC on their computer, or they’re on their sofas, laptop in lap.
- CNN was the most iPhone-ish. HLN was the most Android-ish (but more iPhone than Android). Foxers were slightly more likely to use an iPhone than MSNBCers.
- The web and iPhone are the most popular channels and about equal in usage.
- iPhone beats Android by about 3 to 2 overall.
- Windows Mobile sent about 1% of all tweets. Blackberries sent about 0.6%. To be fair, if you’re using the Microsoft Surface, it probably looks like web.
- Tweetdeck sent about 3% of all tweets, Hootsuite about 0.7%. People are not using “advanced” Twitter clients a whole lot
- The “Mobile Web” (that is, using the web from a mobile device) is tracked separately from “Web” (which, I presume, is non-mobile). And mobile web was about 1% of all tweets. People definitely are using native Twitter clients rather than logging into the web on their mobile devices.
Raw data for your own analysis:
- There were 4,332,044 tweets in total that were counted for this analysis.
- Top 1000 clients across all networks (yes, there are more than 1K): 2014-09 Clients All Networks.csv
- All clients responsible for more than 100 tweets on any network, by network: 2014-09 Clients vs Networks.csv
Caveats:
- The idea of what is and what isn’t a cable news related tweet is not well defined. I have over 150 filters defined to catch the tweets, and I think I do a good job, but whether you think I do is another issue.
- I counted tweets, including retweets. Should I have counted unique senders? Should I have excluded retweets? Each of these is a reasonable suggestion and may well have changed the scoring.
- I could have missed some tweets, although probably not enough to change anything. So I think.