Social influence sometimes creates significant unpredictability in cultural markets (e.g films and music albums) providing one potential explanation why experts routinely fail at predicting commercial success of cultural products. As a result, social influence is often presented in a negative light. We study the dynamics of a trial-offer platform visited by consumers whose choice preferences can be modeled using a multinomial logit (MNL) or a mixed multinomial logit (MMNL). In this marketplace, consumers are influenced by past purchases, the inherent appeal of the products, and their visibility in the platform.
We examine various ranking policies that the platform owner (firm) could use and analyze their computational complexity, their long-term dynamics and predictability. We also analyze whether the effect of social influence is beneficial to the firm. The results obtained differ drastically depending on whether consumers follow the MNL model or the more general MMNL model.
Joint work with Andres Abeliuk, Franco Berbeglia, Manuel Cebrian, Felipe Maldonado and Pascal Van Hentenryck.