Identifying sequential influence in predicting engagement of online social marketing for video games

Abstract

Advancement of online social networks has seen digital marketing use platforms like YouTube and Twitch as key levers for video games marketing. Identifying key influencer factors in these emerging platforms can both deliver better understanding of user behavior in consumption and engagement towards marketing on social platforms and deliver great business value towards video game makers. However, data sparsity and topic maturity has made it difficult to identify user behavior over a sequence of different marketing videos, with a key challenge being identifying key features and distinguishing their contribution to the measure that defines sustained engagement over sequential marketing. This paper presents a method to understand sequential behavioral patterns by extracting features from marketing frameworks and develop a supervised model that takes all the features into consideration to identify the best contributing features to predicting engagement that delivers sustained interest for the next video in a series of marketing videos on YouTube. Experiment results on dataset demonstrate the proposed model is effective within constraint.