Firms, political parties and non-profit organisations increasingly rely on social influencers to spread products, ideas or behaviours. There is extensive evidence that shows that influencers drive new product adoption, public health policies or voting behaviour. Therefore, knowing who these influencers are gives organizations a competitive edge. Identifying them, however, can be a challenging work. If done right, it can lead to an increase in sales and profits or success of social programs. If done wrong, it can lead to a waste of resources on sub-optimal campaigns.

Today, the common identification approach is to decide first on the features that best describe an influencer (e.g. expertise, personality, position in the social network) and then identify the influencers as the individuals with the highest values of these features. In doing so, however, it is often neglected that social influence is both context and time dependent. Social influence is a complex process, that operates through several mechanisms (e.g. contact, socialisation, status competition, social norms) which have a different impact across the five stages of the decision process (knowledge, persuasion, decision, implementation, confirmation). In consequence, an out of the box identification method that works well in one setting (e.g. to identify influencers in political discussions) it not guaranteed to work in another (e.g. to identify influencers that promote a healthy lifestyle).

We studied how the collective wisdom of a social group can be leveraged to identify influencers. We developed a method that does not rely on pre-defined features describing influencers but which considers what is relevant for the analyzed social group at each point in time. Galton (1907) has long ago observed that social groups can make more accurate collective judgements than expert individuals. This phenomenon is known as the wisdom of the crowds (Surowiecki, 2004) and has raised great interest among both researchers and practitioners. The concept has been applied to solve a large variety of problems, from prediction markets to informed policy making and made as well its way into mainstream applications, being an important mechanism behind creating content on social information sites such as Wikipedia, Quora or Stackoverflow.

We applied the principle of the wisdom of the crowds to identify influencers and developed a simple, yet insightful method that aggregates the individual evaluations into a collective judgment. In doing so, we do not pre-impose a set of features describing who the influencers in a social group may be, but we let each individual in the group decide on his own, based on the preferences and beliefs held at that point in time. The aggregation method we developed is computationally scalable and it can thus be applied in real time situations where immediate actions are necessary.

We tested and validated the value of our approach in two distinct contexts. First, we studied news discussions between 2 million users writing more than 50 million posts on CNN, The Atlantic and The Telegraph. We found that by leveraging the wisdom of the crowds, it is straightforward to reveal who is consistently the most influential. We also learned that the extent to which people agree on who are the influencers varies across discussion topics and that most influencers are influential within only one topic, which shows once again that influence is context dependent.

Second, in a different study, we applied these findings to a concrete business situation. We worked together with a manufacturer of kitchen appliances that operates an online community platform where users share their cooking experience. We segmented the entire user base into three groups, identified the influencers withing each group and showed that each type of influencers can contribute to increasing the community value in a different way. Then, we conducted a focus group to better understand the motivation and expectations of the identified influencers and how to leverage their knowledge to create high quality products. We learned that most influencers are not motivated by financial incentives. A potential strategy to increase their engagement is to offer them more responsibility in the community.

Ultimately, the applicability of our research goes beyond influencer identification. The aggregation method provides a highly scalable approach to identify individuals who consistently outperform others in terms of a defined metric that is repeatedly evaluated over time. It can readily be used as well in diverse disciplines like management to quantify performance of employees or sports to identify the most valuable players.


Based on the following papers:

  • Tanase, R., Tessone, C.J. and Algesheimer, R. (2016, June), “The identification of influencer through the wisdom of crowds”, Working Paper.
  • Tanase, R., Tessone, C.J. and Algesheimer, R. (2016, June), “Identifying influential individuals from time-varying social interactions”, Network Science Conference, Seoul, South Korea.

References and further readings:

  • Galton, F. (1907), “Vox populi”, Nature, 75(7): 450–451.
  • Surowiecki, J. (2004), “The Wisdom of Crowds: Why the Many Are Smarter than the Few and How Collective Wisdom Shapes Business, Economies, Societies, and Nations”, Doubleday Books: New York.

Further information about the author: Radu Tanase