Twitter’s spam problem

There’s been an increasing amount of reporting about the underground economy that powers the sale of retweets and followers on Twitter, leading to some questions as to how popular the service truly is. With reports of 500 million registered accounts, what is the actual number of users on the service?

To figure things out, I tried to get a sense of how widespread the problem is and I used the top 25 accounts on Twitter, as reported by Twitter counter:

Name	Account	Number of followers
Justin Bieber	@justbieber	37,156,888
Lady Gaga	@ladygaga	35,866,584
Katy Perry	@katyperry	34,610,671
Barack Obama	@barackobama	29,396,747
Rihanna	@rihanna	29,091,726
YouTube	@youtube	25,885,816
Taylor Swift	@taylorswift13	25,827,240
Britney Spears	@britneyspears	25,424,014
Shakira	@shakira	20,292,896
Justin Timberlake	@jtimberlake	18,463,919
Ellen DeGeneres	@theellenshow	18,146,995
Instagram	@instagram	18,049,505
Oprah Winfrey	@oprah	17,847,895
Kim Kardashian	@kimkardashian	17,613,483
Twitter	@twitter	17,553,144
Christiano Ronaldo	@cristiano	17,272,956
Jennifer Lopez	@jlo	16,451,048
Nicki Minaj	@nickiminaj	16,216,799
Bruno Mars	@brunomars	15,859,677
Kaka	@Kaka	14,908,112
Selena Gomez	@selenagomez	14,684,855
P!nk	@pink	14,630,875
Marshall Mathers	@eminem	14,172,763
Ashton Kutcher	@aplusk	14,088,649
Demetria Lovato	@ddlovato	12,929,805

This provided me with a good sample set of anywhere between 13 and 37 million users, or roughly 2-7% of Twitter’s 500 million registered accounts. Considering Twitter recently reporting that it had 200 million active users, we should be able to observe similar patterns in that data set, or 4-14 million active users.

To assess the percentage of active users, I used a tool called the Fake Follower checker by StatusPeople. When given a Twitter account, the system reports how many users it thinks are fake, how many are inactive accounts (ie. could be real accounts but do not have much activity) and how many are active accounts. Throwing our data sample at this, we got the following percentages:

Account	Fake followers	Inactive	Good
@justbieber	18	24	58
@ladygaga	45	33	22
@katyperry	49	31	20
@barackobama	50	28	22
@rihanna	36	29	35
@youtube	49	27	24
@taylorswift13	44	32	24
@britneyspears	48	32	20
@shakira	44	29	27
@jtimberlake	46	31	23
@theellenshow	44	31	25
@instagram	30	21	49
@oprah	49	31	20
@kimkardashian	35	30	35
@twitter	50	27	23
@cristiano	46	31	23
@jlo	54	32	14
@nickiminaj	31	29	40
@brunomars	35	31	34
@Kaka	52	29	19
@selenagomez	34	31	35
@pink	49	31	20
@eminem	42	31	27
@aplusk	38	40	22
@ddlovato	43	32	25

or, in terms of real numbers, this is what it looks like:

Account	Number of followers	Fake followers	Inactive	Good
@justbieber	37,156,888	6,688,240	8,917,653	21,550,995
@ladygaga	35,866,584	16,139,963	11,835,973	7,890,648
@katyperry	34,610,671	16,959,229	10,729,308	6,922,134
@barackobama	29,396,747	14,698,374	8,231,089	6,467,284
@rihanna	29,091,726	10,473,021	8,436,601	10,182,104
@youtube	25,885,816	12,684,050	6,989,170	6,212,596
@taylorswift13	25,827,240	11,363,986	8,264,717	6,198,538
@britneyspears	25,424,014	12,203,527	8,135,684	5,084,803
@shakira	20,292,896	8,928,874	5,884,940	5,479,082
@jtimberlake	18,463,919	8,493,403	5,723,815	4,246,701
@theellenshow	18,146,995	7,984,678	5,625,568	4,536,749
@instagram	18,049,505	5,414,852	3,790,396	8,844,257
@oprah	17,847,895	8,745,469	5,532,847	3,569,579
@kimkardashian	17,613,483	6,164,719	5,284,045	6,164,719
@twitter	17,553,144	8,776,572	4,739,349	4,037,223
@cristiano	17,272,956	7,945,560	5,354,616	3,972,780
@jlo	16,451,048	8,883,566	5,264,335	2,303,147
@nickiminaj	16,216,799	5,027,208	4,702,872	6,486,720
@brunomars	15,859,677	5,550,887	4,916,500	5,392,290
@kaka	14,908,112	7,752,218	4,323,352	2,832,541
@selenagomez	14,684,855	4,992,851	4,552,305	5,139,699
@pink	14,630,875	7,169,129	4,535,571	2,926,175
@eminem	14,172,763	5,952,560	4,393,557	3,826,646
@aplusk	14,088,649	5,353,687	5,635,460	3,099,503
@ddlovato	12,929,805	5,559,816	4,137,538	3,232,451

Looking through this data, we find that, on average, popular accounts’ twitter followers break down along the following lines: 42.44% fake and 30.12% inactive. This means that only 27.44% of users in that population are active on the service. If we were to look at Twitter’s 500 million registered accounts, this would mean that a bit over 137 million users are active on the service, leaving us with a question mark around the other 63 million active users Twitter is reporting. One could assume that, as you move further down the list of users, the number of inactive accounts drop and the number of fake followers also does. Or could it be that the 63 million missing users in that count represent the population of fake accounts on the service?

We could try to look at it through the lens of inactive users. Twitter is reporting that roughly 40% of their registered accounts are inactive (since they have reported 500 million accounts and 200 million active users). This would mean that, among the top 25 most followed accounts, we have a more active population of followers, as only 30% are reported inactive. However, this also means that there is a 25% divergence between the two numbers. So if we were to apply the same logic to the Fake followers numbers, we may see that 31.83% of Twitter’s users could be fake, or roughly 159 million accounts.

… and if those accounts were acting as active accounts, we could be looking at an active population on Twitter of only roughly 40 million real users.

Why Twitter followers matter

One may think that looking at Twitter followers is a ridiculous endeavor. However, in a world where anything from reputation to credit worthiness is increasingly based on such information, it is important that we look at the source data and ensure its integrity. Today, one can buy Twitter followers by the thousands for mere penny per user. The fake accounts that plague the service are undermining it, creating an environment where one cannot necessarily trust the audience. If Twitter’s future is based on delivering audiences to advertisers, the company needs to do a better job culling its user base and going after those fake follower accounts.

Will Twitter clean up its act? At the current time, it seems hard to tell. The company has not made any indication that it sees the underground economy of follower sales and spam accounts as anything to be too concerned about. If anything, the company may actually be disincentivized from doing anything about it, as those fake followers make the service appear a lot busier than it actually is and appear to be narrowing the gap in user base between Twitter and Facebook.

But as it keeps delaying dealing with the issue, the problem will become larger and more ingrained, making it more difficult to deal with further down the line. Already, fake account creators have now figured ways to create profiles that appear legitimate on first pass. The reason for their working so hard to do so is that better looking fake profiles represent better sales for fake followers vendors (the murky world of followers sales has already divided itself into bottom feeders who will sell any kind of followers and “premium” vendors who claim to provide a guarantee that the followers they add are real.) The market is speaking but it is the same kind of murky market that created the spam that destroyed usenet forums, the spam that made it more difficult to operate email services, and the kind of dark SEO techniques that have gotten the likes of Google and Microsoft to spend millions to fight. Social spam, as it already exists on Twitter, will continue to grow and unless the company addresses the problem quickly, it may be the one thing that sinks it.