An In-Depth Exploration of Twitter

TwitterOver the past five months, Twitter has seen tremendous growth and, in the process, attracted a huge amount of media and blog coverage.

To quantify this growth, Sysomos decided to conduct a study that involved collecting information from 11.5 million Twitter accounts. We believe it’s the most extensive look at Twitter ever done, offering in-depth insight into Twitter’s growth and how people use it.

Among the findings are that:

• 72.5% of all users joining during the first five months of 2009.
• 85.3% of all Twitter users post less than one update/day
• 21% of users have never posted a Tweet
• 93.6% of users have less than 100 followers, while 92.4% follow less than 100 people.
• 5% of Twitter users account for 75% of all activity
• More than 50% of all updates are published using tools – mobile and Web-based – other than TweetDeck is the most popular tool with 19.7% market share.

The complete report on Twitter can be found online on our Web site, as well as a PDF version for download. A complete list of all graphs and charts, in addition to those in the main report, are available in the appendix to the Inside Twitter report.

Twitter: Friends vs. Followers


More: TechCrunch’s Erick Schonfeld did a blog post on the report.

24 Comments on “An In-Depth Exploration of Twitter”

  1. (Disclosure: I’m a paying client of Sysomos).

    Awesome analysis…this report is a must-read and really shows of your capabilities well. How confident are you in the age breakdown since there really isn’t a spot to put age in your Twitter profile?

  2. Great Job! But for me there is still one graph missing: Could you combine the data out of “How old are twitter users” and “What sex have twitter users” to an overall “population pyramid” of the twitter-nation?

    But anyways: great analysis!

  3. @David – Some people disclose their ages in their bio fields, e.g., a 22 yr old web designer from Queens, which is picked up by our parsers (along with many similar variations). However only 0.6% have disclosed their ages of 11.5 million users.

  4. Pingback: Een onderzoek naar Twitter « Doeland’s Blog

  5. Great Report, i was trying to get some similar information through Twitter Api but i couldn’t so far.

    I would love to know the TOTAL users by country for the whole world. Do you think you could publish this information? I’m Argentinian living in Mexico, and none of this countries are listed

    I think this can be very helpfull for SEO and Marketing strategies, in other words, does it make sense to invest in Twitter marketing in Latinamerica? apparently only Brazil “uses” twitter.

  6. Pingback: Erhan Erdoğan » Blog Arşivi » Teknoloji Evangelisti’nden Notlar - 30 Haz 2009

  7. Great report – very interesting.

    I’m curious re: methodology – how did you select the 11.5 M accounts? Why did you choose to analyze that large a number, and not just the entire Twitter population?

    (And on a related note, do you have a “current total number of Twitter users” estimate?)

  8. @Matt We continuously crawl for Twitter data, trying to gather as much data as we can. When we produced the report, during the end of May, we had 11.5M user profiles in our database. While it is impossible for people outside Twitter to know the actual total number, it was rumored to be around 14M at the time. This means our crawlers were over 82% effective in terms of coverage.

    Sorry I don’t have the current number handy, but it can be computed. We plan to release regular updates to the Inside Twitter report with updated statistics.

  9. Pingback: 8 Temmuz 2009 : Google İşletim Sistemi Çıkarırsa? | Ahmet Alp Balkan : Blog

  10. Thanks, great Twitter stats.

    Note that in your summary, you incorrectly use the word less when you should have used the word fewer on several occasions.


    I.e., followers would be a counted item, thus you should say fewer than followers.

    Feel free to delete my comment if you fix the grammatical error.

  11. Thankyou for sharing this interesting data.
    I calculated the distribution of followers count for a (much smaller) sample of users from the public timeline. This distribution looks different to the one you find for all users – which is not surprising, as for a start the public timeline only includes users that tweeted at least once. However, intriguingly, it looks quite similar to the distribution you find for Internet Marketers.

    Could you please tell me how you identified your data set of Internet Marketers? Did it only include marketers who had tweeted at least once?

  12. @Miranda we looked at many user accounts, including the ones without any tweet. The internet marketers were identified based on what was disclosed in their bio.

    If you use the public timeline, this method is inherently biased towards who tweet more. You can not get the uniform random sample that way, and probably the reason for the difference.

  13. This is a fascinating and very comprehensive study. Thank you for sharing.

    I just have a word of caution about drawing causal relationships where you see correlations – for example, you state, “As Twitter users attract more followers, they tend to Tweet more often,” I would argue that as you tweet more often, you are more likely to draw more followers. Logically too it makes more sense to say that the more you tweet the more followers you will have because you are likely to be an interesting person if you have so much to say, instead of the other way round, where I will feel more compelled to tweet because I have more followers now.

    Either way if it is only correlations you are studying, it is hard to make causal statements unless you did other tests that are not reflected here.

    Shalini Bahl

  14. Pingback: Serendipidade Social :: Tutoriais CTDO - Sua Base de Tutoriais Online

  15. Thanks. I am forever very much interested in items and info about website seo. You don’t know where you are going to find a new tactic (well, new to me at least). Although there are always fundamental Search engine optimisation methods, a little tweak here and there, and your blog will receive significantly better search rankings. If you work with WordPress using appropriate plug ins and / or a good premium theme can help your Search engine optimization. It is also really important to keep up to date with the changes in search engine algorithms.All the best

  16. There are many blogs in the internet but yours certainly stands out from many others in the internet. Why do I say that? First, you are regularly posting new pieces of information on your blog. In addition there are many active users reading and commenting on this blog who are willing to share their thoughts regarding the many discussions. Last but not least, the conception of the blog has a certain appearance which honestly speaking I have not found in another blog anywhere in the world wide web.

  17. Alex Cheng and Mark Evans – no sense in pretending there is a baseline here – the comparatives are all going to have to have apples to apples, which isn’t exactly kosher, is it? Why not some analytics with say, Facebook, or, BUZZ, or even…ya know…some other social type networking type construct? Your stats are pretty and the report is pretty, kinda doubt some of your conclusions because I didn’t hear a lot of Null statements with regard to the results, but nice looking work and probably accurate 3-5% about Twitter… ;-);-) wanna shoot me now…;-) Remember what Trochim over at Cornell said…;-)

  18. Thanks much. It’s a terrific report that will be very useful in my class. Kudos to Sysmosos and good marketing….I will remember.

  19. I would like to second the point that Shalini made above that it is not more followers which lead to more tweets but counter clockwise. However, I would also add that we need more data to find a correlation between that. Twitter without doubts is one of the biggest social platform in the making and since its so new, there are millions of ways in which it can be used.

    Great correlative article on the topic though!! Kudos !

Leave a Reply