Bigger isn’t necessarily better: Twitter’s million follower fallacy

In August 2009, Adi Avnit of Dotmad.net wrote a guest post on the Pravda: On Media and Technology blog in which he raised the issue of marketers being seduced by high follower counts on Twitter®. In this post, he argued the case for more robust metrics to determine the "reach" of a given message, and coined the “million follower fallacy” term.

This prompted post-doctoral researcher Meeyoung Cha at the Max Planck Institute for Software Systems to undertake a study which resulted in the paper, “Measuring User Influence in Twitter: The Million Follower Fallacy,” published in May 2010. Fellow researchers were Hamed Haddadi of the University of London’s Royal Veterinary College, Fabricio Benevenuto of the Federal University of Minas Gerais (UFMG) in Brazil, and Krishna P Gummadi of the Max Planck Institute.

Sifting through a mammoth amount of data

The researchers approached Twitter, who "white-listed" some 58 servers, allowing the team to collect and collate information on tweets and users over an eight-month period from the beginning of 2009. Cha et al. provide some interesting statistics, which raise further questions not addressed in the report.

For example, of the 80 million possible IDs, only 54,981,152 represented in-use accounts, suggesting that the other 20 percent were either abandoned during set up or have since been deleted either by users or by Twitter administration.

Those accounts were connected by 1,963,263,821 social links, suggesting the average Twitter user at the time had fewer than 36 connections. Between them, they had amassed 1,755,925,520 tweets “since the early days of the service,” equating to a little under 32 tweets each.

After discarding accounts that were set to private (a little under 8 percent) and those who did not have a valid screen name or who had posted fewer than 10 tweets, “only” 6,189,636 users remained for analysis. The influence of these 6 million was measured, however, against the activity of the 52 million non-private accounts.

The research findings

The analysis focused on three metrics: indegree (the number of followers), retweets (forwarding of one user’s message by other users) and mentions (use of the username preceded by ‘@’). The researchers explained: “Indegree represents popularity of a user; retweets represent the content value of one’s tweets; and mentions represent the name value of a user. Hence, the top users based on the three measures have little overlap.” In fact, only Ashton Kutcher and Puff Daddy figured in all three lists.

The most followed users were mostly news sources such as CNN and The New York Times, politicians (Barack Obama), sports personalities or celebrities (Shaquille O’Neal, Ashton Kutcher and Britney Spears). The most retweeted included content aggregation services such as Mashable, TwitterTips and TweetMeme, businessmen like Guy Kawasaki, and news sites such as The New York Times and The Onion. The most mentioned users, the report found, “were mostly celebrities.”

This supports the researcher’s first finding, that “popular users who have high indegree are not necessarily influential in terms of spawning retweets or mentions.”

The paper’s second assertion, that “most influential users can hold significant influence over a variety of topics,” is based on the analysis of tweets and retweets referring to “the Iranian presidential election, the outbreak of the H1N1 influenza, and the death of Michael Jackson.” These topics were chosen to “span political, health, and social genres.” Although fewer than 2 percent of the users analyzed tweeted across all three subjects, after ranking the top 10 percent of users for each topic for retweets and mentions, a high level of correlation was found across all three.

The third finding, that “influence is not gained spontaneously or accidentally, but through concerted effort such as limiting tweets to a single topic,” is based on a retrospective analysis of the habits of Twitter users who gained in influence over the period studied. The report concluded that, “Finally, we found that influence is not gained spontaneously or accidentally, but through concerted effort. In order to gain and maintain influence, users need to keep great personal involvement.”

In an interview with Brent Lang at theWrap, Cha summed up the findings of the paper, saying, “Follower count is not sufficient to capture influence ... It only shows how popular the user is … Retweets and mentions, which measure the audience responsiveness to a user's tweets, do not correlate strongly with number of followers.”

She also indicated the direction of future research, adding, “this is just the first of a series of studies where we intend to use Twitter data to analyze how information (news, rumors, conventions) spread through out the social network. We expect some of these future studies to go beyond confirming existing theories and yield new (potentially surprising) findings.”

Twitter is a registered trademark of Twitter Inc.