Turns out, that argument you’re having online might not be with a real person. A new study from the University of Southern California and Indiana University finds that up to 48 million Twitter users — or 15 percent of all users — are bots.
In 2014, Twitter estimated that between 5 percent and 8.5 percent of its users were bots, non-human accounts that post, respond to messages, or follow accounts automatically. At best, bots provide a useful service such as news alerts or customer service; at worst they’re associated with spam, harassment, or distorting public opinion.
Twitter has struggled to keep pace with competitors like Facebook, Instagram, and Snapchat, despite the president’s affinity for the service. Even during the election, when regular Trump missives were roiling the news cycle and routinely getting tens of thousands of retweets and likes, Twitter’s user growth slowed, as did its ad revenue.
Estimates from eMarketer suggest that in 2017 Twitter’s total ad revenue will grow to about $2.53 billion, up from $2.5 billion in 2016. More than 10 years after Twitter was founded, the company still hasn’t quite figured out how to turn a profit. It’s lost $2 billion since 2011, according to CNN.
But just because a lot of Twitter “users” are bots, doesn’t necessarily mean they’re bad for users. Though we associated them with spam, automated Twitter accounts perform lots of useful functions, a Twitter told CNBC. “Many bot accounts are extremely beneficial, like those that automatically alert people of natural disasters … or from customer service points of view.”
And it’s true, not all the bots are bad — Google’s “spiders” scour the internet to improve its searchability, Facebook’s “feed fetchers” help refresh your feed on mobile, and some of the bots on Twitter perform useful functions.
They’re also not unique to Twitter. A study released in January by the security firm Imperva estimates that bots make up 52 percent of all web traffic. The internet, it turns out, is composed of more bots than humans.
According to the Association of National Advertisers, which partnered with digital security company White Ops to conduct a study in 2016, bots cost advertisers $7.2 billion in revenue — up a full $1 billion over the prior year’s estimates. On the web, advertisers pay by impression. When a bot sees an ad, it’s eating up an impression that a human was intended to see.
Advertisers pay Twitter when an an action takes place — a retweet, a like, or click-through on a link — so certain types of bots can disrupt Twitter’s business with fraudulent clicks. But if bots are sharing information like regular users, are they necessarily bad for advertisers? It seems that bots, like people, can be good, bad, or neither.