Twitter: Findings on Popularity & Influence

twitter ad model

Focus on retweets, not followers

Gordon MacMillan has posted a good piece on Brand Republic entitled, “Followers Ddon’t Equal Influence on Twitter” that summarizes an excellent piece of research done by M. Cha, H. Haddadi, F. Benevenuto & K. Gummadi, titled, “Measuring User Influence in Twitter: The Million Follower Fallacy.” Excellent research and MacMillan teases out a lot of the top line findings.

Some highlights:

  • Retweets are driven by the content value of a tweet, while @ mentions are driven by the name value of the user (i.e. celebrities)
  • Subtle differences (between popularity and influence) leads to a Twitter inbalance of users who have high follower counts but not necessarily getting many retweets or mentions
  • Many influential tweeters can hold sway over a number of topics, which is interesting as many people on Twitter start in one place and migrate into others

Beyond this, the researchers broke Twitter usage / activity into three distinct areas:

  1. Indegree (# of followers)
  2. Retweets (using RT or via)
  3. Mentions (using @)

Through these distinctions they were able to demonstrate that if you’re looking to spread an idea, maybe somebody like Guy Kawasaki (230,000 followers) might be better than Ashton Kutcher (4.8 million followers).

Their research found that an expert like Kawasaki generated more retweets, while a celebrity like Kutcher generated more mentions, but “fewer than 30% of tweets that were classified as mentions contained any URL…”

A deeper dive into the research reveals some other interesting information. A lot of paper centers of the Gladwell v. Watts debate regarding the role of influencers and the Small World Phenomenon. The paper, at least in my reading, seemed to initially side with Watts, c0nfirming that earlier research (Domingos and Richardson 2001) confirmed that “direct marketing through influentials would not be as profitable as using “network”-based advertising such as collaborative filtering.” The paper later acknowledges that “…it is substantially more effective to target top influentials than to employ a massive number of non-popular users in order to kick start a viral campaign .”

So, what conclusion to draw? I’m not sure, but the paper also referenced Granovetter’s theory of Triadic Closure leading me to believe that the authors were somewhat Watts-ian overall (see my take on Granovetter here.) Until, near the end, when they mention that, “a single most popular influential user spawned up to 20,000 retweets and 50,000 mentions over a random 15 day period.”

The researchers were able to analyse a very large sample set, but a careful read raised alarm bells for me. They started with a user sample size of just under 55 million Twitter users. They then applied a couple of filters:

  1. Active users (at least 10 tweets during users lifetime)
  2. Public account
  3. Valid screen name

That’s a fairly low barrier. From the 55 million they were left with…

6.1 million.

Whoa!? One in nine Twitter accounts is really worth analysing? What does that mean for Twitter’s new ad platform?

Later in the paper the researchers look at three key “world news moments” from 2009: Iranian elections, Swine flu outbreak and the death of Michael Jackson. They note that each of these topics reached a Twitter audience of over 20 million, “indicating that over 40% of users in Twitter were aware of at least one of the three topics.” Now, we’ve all heard (usually old, white) cultural commentators lament the end of the shared cultural experience, usually citing lower television ratings. The shared cultural experience still exists – and I would argue it has expanded – it just doesn’t exist on ABC, CBS or NBC.

As a final concluding thought, the authors show that Twitter accounts that were focused on a single topic had the best opportunity for rapid growth. The ‘perfect storm’ occurs when a Twitter account is focused on a specific topic, thus having established credibility, and then the topic hits the mainstream. From there rapid growth can occur.

Overall, some terrific data and plenty to think about, especially for brands.  An intriguing strategy may be to target multiple accounts with authority that focus on a related topic, rather than trying to chase one high profile celebrity account.