A couple of weeks ago, Twitter’s CEO Dick Costolo admitted that Twitter had a problem–a big, massive troll problem to be exact. Some of the recent problems in gaming circles was on his mind as stories and reports of abuse and bad feelings have been buzzing around. Costolo said that he believed that this sort of behavior and Twitter’s lack control over this sort of behavior was driving away core users and hampering Twitter’s growth.
Twitter’s growth has probably been on Costolo’s mind a lot lately also given the degree of attention it has been receiving in the media. Even though Twitter’s financials are doing pretty well and its advertising changes are being well received, Twitter is still just a drop in the bucket compared to Facebook and its massive user base. But ugly, angry trolls aren’t the only problem that Twitter is having–it’s also having bot problems.
A recent study has concluded that around 67% of Taylor Swifts users are bots. Bots in this case are referring to accounts that are believed to be ‘fake’ accounts, but can also be referring to accounts are no longer active–perhaps ‘zombie’ accounts? Twitter only lays claim to having about 5% of its accounts being bots, but frankly I’m not so sure.
A Personal Example
I’ve been toying around with some shameless “personal brand” promotion partly out of intrigue for the last couple months and my results have been rather interesting. This is by no means a scientific report, but it may prove to be a rather interesting look at what is actually going on with Twitter.
According to most social media marketing blogs one of the keys to success with social media marketing is engagement–after all its called social for a reason. So we are talking about tweeting to people, retweeting other people, and following people. I have limited time–and interest–so I decided to try my hardest to follow people. I tried to do this using two different tools.
The first was Twitter’s own ‘who to follow’. Twitter is endlessly nagging you to follow new people. They put it on the sidebar, they pop it up after you follow someone, they will email suggestions to you, etc. It is really in their best interest because they want you to feel engaged and if they can succeed there, then they are more likely to retain you as a user. The other tool I used was Klout.com. Their purpose is to help you find people to follow who are relevant. People who follow your interests, are “influential” in your interest, and are actively promoting content within your interest.
So I followed away until I eventually ended up following 2,000 people and …boom! Did you know that twitter has a limit to how many people you can follow? From the horse’s mouth:
Twitter’s technical follow limits:
1. Every account can follow 2,000 users total. Once you’ve followed 2,000 users, there are limits to the number of additional users you can follow. This number is different for each account and is based on your ratio of followers to following; this ratio is not published. Follow limits cannot be lifted by Twitter and everyone is subject to limits, even high profile and API accounts.
2. Every Twitter account is technically unable to follow more than 1,000 users per day, in addition to the account-based limits above. Please note that this is just a technical limit to prevent egregious abuse from spam accounts.
3.Accounts are also prohibited from aggressively following other users. Our Follow Limits and Best Practices Page has more information on Twitter’s following rules. source
Catch that? The interesting bit I think is right after the obvious problem I ran into–following 2000 people, without a ton of followers myself–you can’t technically follow more than 1,000 users per day. That’s a lot of following! Which means that there is a lot of leash room with Twitter’s policing. But even with limits like that, do they use it?
So lets say that average joe me decides that 2000 is too many to follow and he starts cutting back (which is what I decided after my rather rude awakening by Twitter’s following policy). That won’t affect his following will it? mmmmmmm….think again. Check out the following graph:
You can’t really tell the dates from it, but the top of the graph occurs around 1/20/2015. I had roughly 800 followers and some change, and I was following 2000 accounts. The new low is current, 2/16/2015, and I now have roughly 400 followers and I’m following 694 accounts. So how do we account for radical drop-off in followers?
The only reasonable expectation is that of the 2000 accounts that I was following 400 had followed me back–or followed me first and I followed back, like a good amateur twitter social media marketer! But when I unfollowed, to get my news feed back into a manageable shape, suddenly those 400 accounts decided to hit the road. In short, those 400 accounts had to have been either bots or highly automated marketing accounts that unfollow when you unfollow–in other words bots!
No I realize that there is a technical difference, but is there really a difference? Either way, you are dealing with accounts that are not in the network to engage and enjoy, they are there to do exactly what I was doing–marketing. (Should we call it a social marketing network, instead of a social media network) Now, I’m not complaining, because I was just as guilty of this as they were, but it does serve to illustrate the way that ‘micro-blogging’ network is being used and the magnitude of the behavior.
So What’s the Point?
The moral of this lovely little anecdote is that twitter must have a substantial bot problem, and I highly doubt it is the numbers quoted in the study mentioned before. If an average joe followers consists of nearly 50% bots, what about an account like Taylor Swift or Barack Obama or Tim Tebow? Accounts with that many followers–influencers, like Klout likes to identify–are probably being targeted for following by bots right and left. If this is the case, I suspect that Twitter’s new advertising plans may be the best way for Twitter to find out who is and who isn’t a bot based on user interaction and not just follows. That should give Costolo something else to worry about.