Detecting Influencers using Machine Learning

In the western world, customer have mostly outgrown the phase where they idolize celebrities and have the urge to buy any product that celebrities use. Now, customers are looking for communities of people who share their values and make them feel like they belong. These communities are not built around just anyone though. The profession of influencer marketing has been created whereby a specific type of person is at the center of an online community and wields a great deal of influence power on their peers in the community. These influencers have become a type of mini-celebrity; they are famous enough to be looked up to but not so famous as to be unapproachable.

Marketers have thus turned to these social media influencers in order to have them introduce the brand to their community. Companies can, for example, send their product to the influencer who will then post their review online to their community. The goal is to find a social media user who has a large enough following and a very high engagement in order to get the best ROI. With the advent of big data, this goal can be optimized through social network analysis. The idea is to run greedy algorithms that test each nodes effect on the entire network by the end of the relevant period. These algorithms are improving at such a rate that they are almost to the point where they can predict the emergence of a market influencer before that social media user has even reached that stage. This allows companies to be able to target them first before the other companies in order to form a strong relationship (at a smaller fee, since this influencer is not yet being contacted by other companies). To get an idea of the two basic influence models and to get an introduction on how the algorithm works, feel free to read my short paper.

This Post Has One Comment

  1. Tania

    Great article! Thank you 🙂

Leave a Reply