12 November 2012

Twitter sniff test for fake followers (read: @GregoryDEvans)

There are a number of ways to look at someone's Twitter account to see if they have artificially inflated their followers (i.e., buying follower lists), but I wanted to add one of my own. This isn't a definite test, but should give you a good sense of whether someone's followers are legit.

It's a very simple ratio of followers to lists. In the case of my account, it's 2,295 (as of 11/12/12). Now figure the number of lists the account is on (this isn't displayed directly by number on Twitter, but is displayed on Tweetdeck or other similar Twitter clients). Again, on my account, it's 141 lists. So 2,295:141 reduces to about 16.3:1. In my experience, I have found that a range of 10-25:1 is very common among most people. I don't have enough data to know how well this ratio scales to significantly higher number of followers, but the ratios to tend to be higher; in the range of 40-100:1.

The idea here is that your legitimate, engaged followers are likely to put you on their lists, while fake followers don't actually do anything.

For this blog post, I looked at (an admittedly small sample of) ten people I follow, who have followers from under 100 to over 100,000. In all but one case, the ratio was between 10-25 to 1 (in the one off case, it was 6:1, an argument for more engaged followers, not less).

Now let's take a look at one example, the infamous @GregoryDEvans. He has 42,331 followers but is only on 45 lists. A ratio of 940:1! Realistically, he should be on 2,500-3,000+ lists (and even here, he is on several lists that are unique to his situation). Fake followers, anyone?

Feel free to punch holes in this test. Any ideas?
Post a Comment