A Multi-generational Diffusion Model with Social Media Effects and Pre-Launch Forecasting of Smart Phone

This paper proposes adiffusion model for multi-generational smart phone by using social media which allows us to forecast sales not only at early stage after launch of specific generation but also before its launch. The model is based on multigenerational generalized Bass model and includes the hierarchical model on the structure of parameters for connecting sequential generations. The social media topics are extracted by the labelled dynamic topic model and they are plugged in the adoption rate function and hierarchical model for parameters as covariates. The model reveals how social media accelerates and decelerates the diffusion in the pre-launch and post-launch phases. Unlike previous multi- generational diffusion models, this model forecasts sales of new-generation products before launch. The empirical results show that the model forecasts the unlaunched product sales with better precision compared to extensive comparative models including sentiment analysis and non-diffusion model, and social media topics accelerate sales in the pre-launch period and their effects decrease with varying patterns in the post-launch period. In conjunction with its effects on switching and leapfrogging, the model provides useful information for product management over long range time horizon.