New Study Finds 66% of Tweeted Links to Popular Websites are Shared by Bots
If you’re still using Twitter data – like Likes, re-tweets and shares – as indicators of success for your social marketing program, you likely need to re-assess your process.
As has been widely covered in recent times, Twitter’s systems can easily be gamed through purchased bots and interactions, which largely de-values such metrics.
The latest findings on this front come from the team at Pew Research – after conducting a study of 2,315 of the most popular websites, and 1.2 million tweets (sent by English language users), Pew found that some 66% of all tweeted links to these sites are “shared by accounts with characteristics common among automated bots, rather than human users”.
That’s not to say all bot accounts are no good – some have a clear purpose, as pointed out by Pew, including accounts like @netflix_bot which automatically tweets when new content has been added to the online streaming service.
But many of these bot accounts are simply inflating numbers – if you see an account with a lot of retweets, that doesn’t necessarily mean their audience is actively engaging with that content.
As noted, the study comes after a recent New York Times report which found that a large number of high profile Twitter accounts have purchased followers and retweets in an effort to boost their apparent popularity. Some of these users are then using that perceived status to frame themselves as ‘influencers’, and subsequently being paid to promote products, largely to automated bots.
This also comes after researchers found huge networks of Twitter bots last year - some numbering up to 350,000 automated accounts – which had been used, at different times, to influence the popularity of tweeted topics.
For their part, Twitter is taking action. In recent times, they’ve been removing bots at a higher rate than ever before, while they’re also looking to make changes to their API to limit misuse (though those changes on hold right now). But even with those actions, Pew’s findings do underline that Twitter interaction data can be misleading, and that people need to look a little deeper to ensure that Twitter users are actually seeing the engagement their numbers suggest.
So how can you do this? The simplest way is by looking through their actual tweets – if a user has a million followers and hardly any interaction on their tweets, you can probably assume they don’t carry the influence the figures might indicate.
You can also use a tool like Twitonomy, which will give you additional stats on a users’ interaction and activity, while an app like Twitter Counter will show you if a user has seen any sudden jumps or drops in follower counts, which likely suggests they’ve purchased followers.
But the key lies in your own analytics data. If you’re seeing interaction on Twitter, but you’re not seeing subsequent click-throughs, it’s probably an indicator that the platform isn't generating the results you want. Twitter activity can still be a good marker for awareness, and there are still uses for such data, within context. But it is worth noting the findings of these studies, and considering which Twitter metrics are actually relevant to you. And then, how to best measure them.
Follow Andrew Hutchinson on Twitter