A targeted approach to viral marketing

•A targeted approach to viral marketing is proposed.•This approach exploits knowledge about potential market to improve effectiveness.•Local centrality scores and their application to viral marketing are investigated.•The approach is more effective and leads to more activation then the global one.•This approach adjusts to different target markets by showing diversity. Viral marketing can be an effective marketing technique in social networks. Initiating from a set of influential seed users, it can activate a “chain-reaction” driven by word-of-mouth. The effectiveness of viral marketing lies in the fact that it conveys an implied endorsement from social ties. Existing approaches to selecting influential seeds depend on measures of global centrality within the structure of the social network – they select users that are central in the entire network according to some centrality measure (e.g., Eigenvector centrality). In this paper a new targeted approach to viral marketing is proposed that exploits prior knowledge about the potential market and uses local centrality scores to identify seeds that have high chances of reaching and activating many users in the potential market. The performance gained by the proposed approach is investigated with an experimental evaluation that uses data from real social networks. The results show that targeted approach outperforms existing, global centrality based methods. It is also shown that the relative performance of the targeted approach improves in the case where the majority of users are indifferent (or negative) to the viral marketing campaign.