Social Media Analytics and Business Performance: a Predictive Modelling Approach

Organisations collect social media metrics, including engagement, sentiment, reach and mentions, but seldom connect them to business performance in a way that supports decisions. This paper develops a predictive modelling approach in which lagged social media metrics are used to forecast weekly sales, and evaluates it in a simulation of 156 weeks in which sales respond to engagement with a two-week lag, to sentiment with a one-week lag and to mentions with a three-week lag, alongside reach and paid promotion. A ridge regression on lagged social metrics attains an out-of-sample R² of 0.73 and a root-mean-square error of 10.8, against 0.07 and 19.9 for a baseline on lagged sales, paid promotion and reach alone. Permutation importance identifies engagement lagged two weeks and sentiment lagged one week as the dominant predictors, recovering the true lead structure. The paper sets out the approach as a four-step method, metric selection, lag structure estimation, regularised modelling and out-of-time validation, and discusses how its outputs support marketing allocation, early warning and the attribution of social activity to performance. It also identifies the conditions, including metric reliability, stable lead relationships and adequate history, on which the approach depends.