How Many Fake Followers Does Katy Perry Really Have?
By Rob | 22 Nov 2013, 12:30pm | Category: Social Media
It's Friday and we think it is appropriate we deal with one of the big issues of the day... How many fake followers does Katy Perry really have?
Answer? Quite a lot -- probably.
Let me explain...
As people who use our Fakers App know, it is very good at providing results for regular Twitter users and brands. Those with less than 50k followers. Which according to our figures is about 95% of Twitter users. But once you start looking at the outliers, the Katy Perrys and Barack Obamas of this world our app is less accurate. Or at least that is what we thought.
In recent months a number of people such as Tony Harris from @D3liberate and @JeffElder of Wall Street Journal fame have been asking us if we can provide more accurate scores for bigger Twitter accounts. And well, that is exactly what we have been doing. We're calling it Deep Dive as it allows you to dive far deeper into a user's Twitter Follower data than ever before. It's currently in test -- live soon -- but is producing some scores. Which, of course, includes Katy Perry's.
So what have we learnt about Katy Perry? A lot, and we're quite surprised by the results.
Currently our our app, which will assess the first 50k followers of any Twitter user, returns a score of 39% Fake, 50% Inactive and 11% Good for Katy Perry. By contrast Deep Dive samples the first 1.25 million records of any Twitter account.
Now, my assumption before beginning this work was that the fake followers of a big account would drop off significantly once you could sample more than the first 100,000 follower records. My belief was that spam would be concentrated in the most recent data. And that the older data would be cleaner and more legitimate because Twitter would have spotted and removed the older spam and fake accounts.
On this point I am probably wrong. The data suggests the issue is worse than I thought. So for Katy Perry we managed to assess 33,000 followers across her first 1.25 million followers. This is a very large sample of data. And surprisingly the scores came back 42% Fake, 36% Inactive and 22% Good.
So in essence there is little difference between our App's score and the Deep Dive score. Yes their are less inactive followers, but it is not a significant change and we've found a few more fake followers.
My conclusions from this are as follows. Spam and Fake accounts on Twitter are probably a bigger issue than I thought. Twitter could possibly do more to remove these fake accounts. Katy Perry and other celebrity figures are probably not as popular on Twitter as the follower numbers suggest.
Overall I think this data suggests there needs to be a wider discussion about follower quality on Twitter. And brands and consumers need to demand more from Twitter on this subject. Because ultimately this is an issue of market legitimacy.
- We're not claiming 100% accuracy on the above as Deep Dive is still in test, but we are confident in the scores.
- The scores were generated without using our most recent algorithm.
- We are happy to share the data we have on this with Twitter if they would like to see it and validate it.
- Picture courtesy of The Hollywood Reporter