Dive Brief:

• A joint study titled "Online Human-Bot Interactions: Detection, Estimation, and Characterization" from the University of Southern California and Indiana University found that up to 48 million — or 15% — of Twitter accounts might be bots rather than people, as reported by CNBC.
• The researchers used more than 1,000 features such as friends, tweet content and sentiment and time between tweets to uncover bot accounts, estimating that the figure lies somewhere between 9% and 15% of total active Twitter accounts. The report also noted that, even given the research parameters, some complex bots still might have been mistaken as human, so the 15% tally could be a conservative estimate.
• The report’s conclusion made the point that most social media platforms make it easy for accounts "controlled by hybrid or automated approaches to create content and interact with other accounts."

Dive Insight:

The universities' findings, if accurate, could hardly come at a worse time for Twitter, which recently saw advertising revenues decline for the first time since going public.

For its Q4 earnings report, the struggling social media company touted a strong, dedicated user base, but a widespread infestation of bots potentially damages that argument and could cause marketers to pay even less attention to the platform, which many brands only use as a customer service tool anyway. Twitter currently has 319 million monthly active users, and its audience growth has been relatively stagnant of late. 

The report did point out, however, that not all bots are problematic because many do exist to perform customer service functions, send alerts or distribute news. In November last year, Twitter launched its first two dedicated chatbot products for businesses, called Welcome Messages and Quick replies. 

At the same time, many bots are malicious and others have been traced to terrorism recruitment and propaganda as well as faked grassroots political support.

Twitter’s own estimation of bot accounts is lower than the study, with an SEC filing from last month indicating that 8.5% of active accounts exhibit bot characteristics when contacting Twitter servers.