In this post, Gal Oestreicher-Singer, Hilah Geva, and Maytal Saar-Tsechansky, discuss the extent to which users of twitter use the platform in order to diversify their identities, or to maintain ‘on message’ branded identities. Presenting a novel methodology, their findings suggest that twitter has become a tool for targeted self promotion, behaviour that is especially prevalent in professional bloggers.
More than 20 years ago, in a now-classic manifesto, Tom Peters announced to the world that branding was no longer the domain of companies alone: regular people must also become marketers for their own “brands” — “the brand called You” — and actively ensure that the world perceives them as they wish to be perceived. That once-novel idea has become an integral part of the way we live now, in large part thanks to social media, where a seven-year-old playing with toys can gain millions of YouTube subscribers, and a corgi can become an Instagram influencer. Even without approaching that level of social media celebrity, most people have engaged in some form of impression management online, be it curating a profile picture or agonising over the wording of a post, hoping to generate a satisfactory number of “likes”.
Importantly, social media provides more than just a channel for disseminating personal brand content: it provides the content itself. In fact, it is possible to develop a social media presence without expressing a single original idea, simply by re-posting content created by others. This means that, with relatively minimal effort, users can use social media platforms to reinvent themselves. If you’ve always wanted to be seen as a sports expert or a music buff, you don’t have to invest the time to become knowledgeable in those areas; you can just share content produced by people who already are. The question is, do people take advantage of this opportunity? Or to put it differently, how do social media users incorporate others’ content into their online personal brands? Do they use it to create personas that go beyond the boundaries of their own expertise? Or do they stick with what they know, sharing content related to domains they are already familiar with? We set out to answer these questions.
Tackling Twitter data
We studied data from 3,562 active users on Twitter and the users whom they followed (their “followings”), collected over the course of six months in 2016. On Twitter, a user can either post (tweet) her own, self-generated content, or re-post (retweet) content generated by others, which the user observes in her “home timeline”, or “feed”. We started by assuming that a person’s brand, or persona, on Twitter can be captured in the topics that he or she discusses. Then, we asked: When people retweet content on Twitter, do they tend to introduce new topics that they don’t typically address in their self-produced tweets? Or do they retweet content related to topics that they already discuss?
The approach to answering our questions might seem simple: Compare users’ retweets with the tweets that they themselves produce, and see how similar they are in terms of the topics addressed. However, there might be a variety of factors — beyond personal branding considerations — that might cause a user’s retweets to follow a certain pattern. These factors must be taken into account before we can fully understand Twitter users’ persona-building behaviour. Here are a few of the points we considered:
Exposure bias. Obviously, a person can only retweet content that she sees. Most of that content comes from the people whom the user follows, and most users don’t choose their followings at random. In particular, people tend to follow others who are similar to themselves. This means that the topics a user is exposed to are likely to be similar to the topics that he or she already tweets about, which means that users choose their retweets from a biased pool of content.
Need for uniqueness. Consumer research shows that people want to stand out from the crowd, especially on social media. This means that users tend to share content that introduces new information that their followers are not likely to have seen before.
Social considerations. Because Twitter is a social platform, users’ tweets are expected to reflect various well-known social behaviours. For example, a Twitter user is likely to show reciprocity, that is, to retweet content posted by others who have retweeted her content in the past. Moreover, users may tend to retweet content produced by more popular users, or by people outside their close social circles.
Constructing and analysing each user’s “baseline persona”
To address our questions, we compared each user’s actual retweets to the retweets of a fictitious “baseline persona” — a persona that represents what the specific user would have retweeted had he posted the same number of retweets that he actually posted, but selected tweets at random from his feed. In constructing this persona, we took into account each of the considerations outlined above (in addition to others). For example, to control for a given user’s need for uniqueness, we ensured that the retweets included in his fictitious persona matched his actual retweets in terms of their popularity on Twitter. Thus, if the user only retweeted items that had been retweeted by others fewer than ten times, his baseline persona would include only items that had been retweeted by others fewer than ten times.
Then, we used an approach called Latent Dirichlet Allocation (LDA) topic modelling to analyse the content of users’ actual personas and their corresponding baseline personas. This approach enabled us to identify the number of topics discussed in each user’s actual retweets, and to compare it with the number of topics discussed in the retweets of his corresponding baseline persona.
A whole new me? Or an upgraded version of me?
We found that people tend to retweet content that resembles the content that they themselves discuss, rather than to introduce new topics. For example, according to our analysis, we would expect a typical user retweeting randomly from the content she observes (while taking into account the factors outlined above) to introduce between 3.74 and 5.80 new topics, whereas in fact, users added less than half that: between 1.94 and 2.45 topics. When we focused on 464 professional bloggers, who are particularly likely to use Twitter as a personal branding tool, we observed that they added even fewer new topics. Notably, these behaviours are in line with established branding strategies, in which companies attempt to portray themselves as “consistent” and “authentic” in an effort to make consumers more receptive to their messages.
Our findings provide new insights about how people use Twitter, and the types of content they are likely to retweet. These insights are likely to have profound implications for firms’ social media strategies. Moreover, our notion of the baseline social media persona, coupled with LDA topic modelling, provides a new means by which researchers and practitioners can analyse social media data, towards understanding what Tom Peters called our “new brand world”.
This post is based on the authors’ co-authored paper, paper Using Retweets When Shaping Our Online Persona: A Topic Modeling Approach, MIS Quarterly, forthcoming and originally appeared on the LSE Business Review.
Note: This article gives the views of the author, and not the position of the LSE Impact Blog, nor of the London School of Economics. Please review our comments policy if you have any concerns on posting a comment below.
About the authors
Hilah Geva is a PhD student at the Coller School of Management at Tel Aviv University. This work is part of her dissertation research.
Gal Oestreicher–Singer is a professor at the Coller School of Management at Tel Aviv University.
Maytal Saar-Tsechansky is a professor at McCombs School of Business at the University of Texas at Austin.