Clicky

10 years of Twitter growing


10 years of Twitter growing


From this report, you will learn how many users (or more accurately, accounts) had Twitter on February 1, 2016, and why no one can say exactly the number of users at the moment. No one, except for the Twitter, of course.


A social network sells its own audience to advertisers and sells to investors a future cash-flow from advertisers.


Advertising prices and equity estimations based on a number of monthly active users (MAU). For a social network, any attempt to join is very important, because it can be a future active user. The conversion rate “new users to MAU” is the key metric for quality of digital audience management.


None of a social networks disclosure the data about monthly joined users and conversion rates. There are no forms for SEC filings for that cases. There are no any official quarterly reports, about the structure of quarterly valuable audience: when were joined users, which brings 80% of current quarter advertising revenue.


Management of any social networks keeps silence about its conversion rate. There are no official quarterly reports about the structure of dropouts: when was joined accounts, which are 80% of the quarterly inactive audience.


The metrics for dropouts and for a most valuable audience are the equivalent of financial statements. Any audience which selling to advertisers is an asset. Any social networks accounts which are inactive is the liabilities. Monthly audience activity balance sheet is a key performance indicator for any digital audience assets.


The core business of any social network is a resale of human's digital attention. Monthly audience activity balance sheet is the major statements for corporate governance. It shows the answer, how effective management spending money on marketing, development, and M&A. Why will someone invest in attention, which is an illusion? Investors need an independent meta-data analysis for audience verification.


Twitter management never published the official data of new joining accounts.



Audience verification company Socialpuncher.com did the independent examination of Twitter. The verification research based on 5,000,000,000 twitter requests.


Socialpuncher.com team made Twitter API requests for the every of ID from 1 to 5,000,000,000. and repeated this procedure 2 months later (so, we can talk about 10B requests). The team has used free Twitter API keys (a lot of them), our own Twitter API interface (ID data crawler) and own data analytic system.


1,426,616,402 from 5,000,000,000, id was successfully checked as really existed until 19 February 2016. All they can be tracked by id requests.


Why can’t we now count the total number of users? Any registration begins with the assignment of ID. If the attempt is unsuccessful, the ID temporarily or permanently remains is reserved. By using an automated mass registration, a very small percentage of the attempt is successful.


Fourth billion of IDs was expended over 9 months (January-October 2015), the fifth billion was expended for 3 months (from 24th October to 2015 19th February 2016).


From 2 February, Twitter began to use the 18-digit random IDs in parallel with ordinal IDs. Since February 19, ordinal ID of the new accounts has been abolished. New accounts can be detected only if they commit any action, for example, follow or retweet someone.


The first eighteen digits ID in our database was registered 2nd February 2016:

ID 694221046598230000, @Valeria01515712, http://archive.ec/ZZ3yk



The target of the research


Independent analysis of Twitter growth dynamics and highlighting the anomalies. The research will be detailed by audience segments.



The key findings of the research


Every quarter from the 4th quarter of 2011 Twitter growing by more than 50 million accounts.


703,000 new accounts per a day joined Twitter for 4 years ( 2011-2015).


Since the 2-th quarter, the audience stagnated, according to official figures has grown within the statistical error.


Half a billion total users Twitter has reached in June 2012. There was the study in July 2012, devoted to that.


Billion total users Twitter has reached in June 2014.


One and a half billion total users Twitter should be reached in May 2016.


If all its registered users would be monthly active, Twitter would have become Facebook.


Of course, the most interesting is the data on the years 2015-2016, when Twitter is experiencing serious difficulties with the growth of the audience, according to official stats. After the 1st quarter of 2015, when it gained 14 million active users and crossed the psychologically important level of 300 million users, the growth has almost stopped.


Over the next 9 months of 2015 joined 171 million new accounts. In 2016, at the moment we have tracked more than 134 million new accounts, registered in 2016, though their numbers are definitely more. Over the past 18 months, with almost stagnating MAU, Twitter user base has increased by 305 million new accounts, but only 11 million become active. Conversions within the last 18 months amounted to only 3%. Although that it is a key indicator of the growth of any audience, it has no place in official reports.


The highest percentage of conversion was in the first and second quarter of 2011, 38% and 39% respectively. The following 3 years it has gradually declined to 20%, fell to 6% in Q1, 2014.


The most inexplicable look stats of the 4th quarter of 2015. We can see 2 million reductions of the active audience against the backdrop of 61 million new registered users.


Explain this and other anomalies is impossible as long as we do not know exactly which 313 million accounts (name, ID) from more than half a billion Twitter considers the monthly active. Even more interesting would be to get an accurate list of active accounts of the 1st quarter 2016 and compare it with the previous and the following quarters. In fact, this is the only step that would make it possible to understand what is really happening with the audience of Twitter.




The appendix





















Twitter is full of anomalies


During the total checking of ID, Socialpuncher research team has found the 168,000,000 of reserved IDs in the one range and 3M stealth botnet, created with reserved IDs. Investigation about 3 million accounts which have been registered on the same day, April 17, 2014, you can read on Sadbottrue.com Since its creation protected botnet made 2.6 billion tweets (including retweets). This is not the only anomaly. We will continue to describe them in the following articles of our study.



Examples of close to first and following billion ID accounts


ID User name Joined Twitter Archive link

1,000,000,004

@VerlWalkingElk

10 December, 2012

http://archive.ec/piTqe

2,000,000,000

@sfa_2000000000

21 March, 2014

http://archive.ec/eNYk5

3,000,000,000

@IsidroQuezad

29 January, 2015

http://archive.ec/p2Pmz

4,000,000,034

@sheenarosefall1

24 October, 2015

http://archive.ec/ooId

4,928,755,216

@Amie98543667

19 February, 2016

http://archive.ec/I7sDp



Among fifth billion of IDs, only 8,5% (85,668,290) was successful. The last ID of registered user is 4,928,755,216.



The background of the research


In 2015 Socialpancher research team have bought 5000 twitter bots for an audience verification experiment. The result of the analysis of bots graph of connections was compiled “Top-100 twitter bot buyers” and discovered the whole botnet of 19,000,000 primitive bot-followers. All the results:Top-100 twitter bot buyers


Based on definitely bot`s behavior patterns and the study of the control group of 2000 definitely human’s accounts, Socialpuncher research team have developed the methodology “What is “bot”?” The methodology is available on socialpuncher.com/methodology/what-is-bots/


Based on methodology, was studied “Twitter top 100”. All 100 accounts have a total of 2.8 billion followers (the exact amount is 2,833,637,341). Twitter Top 100 has in total 523 million unique followers. If all who follow at least for one member of Top 100, would follow a single account, for example, @TwitterTop100, this giant account would have precisely 523 million. The whole “Alternative Twitter top 100” is available on http://socialpuncher.com/top-100/



The key findings of research of 523 million unique Twitter top 100 followers


The total number of unique level 1 bots is 310,537,272.

59,33% of Twitter Top-100 audience is a cheap poor-quality imitation.


How Many Primitive Bots Follow Top-100?



The key question of the research


Twitter`s official metric “Monthly active user” does not specify, is it human, or is it bot.


How many bots are among official 311,000,000 Twitter’s MAU?