First, nobody but Twitter is capable of answering this question, simply because of the fact that we do not have any information about accounts who do not send tweets, but who do log in to Twitter and only consume tweets. Actually if we extend the definition of a Twitter user to these passive users, Twitter themselves may not have the exact answer either, since the impact of Twitter is everywhere. Many websites are embedding tweets on their pages and tweets are constantly read and interpreted by all news media in the world, which incorporate them in the communication with their own audience.
Secondly, determining active accounts is also a matter of definition. Does one take into account users that have submitted a tweet in the last week, month or even longer period? One could also argue that just checking the last tweet sent is not enough to be classified as an active Twitter user, but that someone should send tweets on a regular basis to qualify as an active Twitter user. Obviously the definition of “regular” in this context is also open for debate.
If we start with very broad definitions of active Twitter users, we estimate that 50 million Twitter accounts submitted a tweet less than a week ago, 74 million did this in the last 30 days and 94 million Twitter accounts have submitted at least 1 tweet in 2012, so in the last 2 months.
Checking for the last submitted tweet does not seem to be a very satisfactory way of measuring active Twitter users since we estimate that there are 59 million accounts that have only submitted 1 tweet ever (and 192 million existing accounts that never sent a tweet). When we qualify accounts for inclusion in the Twopcharts we consider accounts active that send on average at least 1 tweet per week, and have submitted at least 1 tweet in the last 30 days. When we apply this definition to the sample we took from the total Twitter population, 51 million Twitter accounts can be considered active as of the end of February. Stretching the criteria to 1 tweet per two weeks and the last tweet not sent more than 60 days ago, the number would increase to 67 million active Twitter accounts.
In a later blog post we will give some detailed analysis of the 51 million active Twitter users, according to our definition.
Some more interesting facts from the data:
- 13.9% of all registered accounts are inactive. These can be both deleted accounts and accounts suspended by Twitter. This leaves a total of 439 million existing accounts.
- 18.8%, or 83 million existing accounts have never sent a tweet and are not following anyone, which make the likelihood that these are inactive accounts very high.
- Out of all of the existing accounts, 29 million, or 6.7% are protected accounts.
- 48.3%, or almost half of the existing accounts are still using the standard egg as their profile picture.
- 21.3% of the existing accounts have submitted a description of the account, 26.6% have submitted location data, and 16.4% have submitted both.
Twitter ‘s official page shows that there are more than 100 million active users and that there are over 250 million tweets sent per day. How does this reconcile with our numbers?
From the numbers above we can conclude that there are 356 million Twitter accounts that may be active in one way or another. This 356 million is made up of the total existing 439 million accounts minus the 83 million accounts that are not following anyone and have not sent a single tweet. It is possible that from the 305 million accounts that are not active according to our definitions, Twitter does consider a part as active, because they are passive readers of tweets, or simply because they use another definition than we do.
With respect to the amount of tweets sent we can try to make an estimate from our data, because we know that the active Twitter users send 4.3 tweets per day on average. If we multiply this with the estimated number of 51 million active Twitter users, we can calculate that at least 222 million tweets are sent per day, which is 11% lower than the latest number as communicated by Twitter.
How did we get these numbers?
We took 10 substantial samples of random Twitter id’s, did some queries on the data, and checked how homogenous the results were. After we were satisfied that the variations between the samples were small, we brought them together in one large sample and considered this sample as a representative sample of the total Twitter population.