Social Value: A Computational Model for Measuring Influence on Purchases and Actions for Individuals and Systems

Measuring influence of one person on another has applications in advertising and marketing and across the sciences. Most approaches involve inferring influence based on speech and social media. In contrast, this paper takes existing spending data and attributes influence on to the spenders and those likely to have caused their spending. The resulting metric, Social Value, is expressed in units of behavior over time. While a person’s total influence on others is called their Social Value, a person’s behavior caused by someone else is called their Following Value. These metrics can also be used across an entire community, customer base, or audience, allowing an objective measure of how much spending or other behavior is social versus nonsocial. These measures in turn open up the potential to test interventions and campaigns to measure viral spread as well as overall shifts in social influence. This article presents a computational model for estimating Social Value, as well as validation of the estimation approach in a study involving players of an online game. A noncommercial open-source implementation of the computational model accompanies this paper.