Measuring Relative Impact of KPIs – A Game Theory Approach

Abstract

Customer satisfaction (CSAT) and Net Promoters Score (NPS) are related to key measures of future financial performance for firms. The ability to find key drivers for CSAT & NPS is an important step in any organizations’ strategy that leads to high quality of service and long-term relationship with customers. One of the traditional techniques to assess the relative importance of attributes is using standardized regression coefficients. But the presence of high degree of multicollinearity among the attributes leads to imprecise and unstable coefficients and thus assessing relative importance among predictors is a challenging task. To handle this type of problem, Shapley value regression (which was developed by Nobel Prize winner Lloyd S. Shapley), which is a cooperative game theory concept using R-square decomposition is applied. In this method, the R-squared is decomposed into contributions from the different predictors in the model and these contributions are referred as importance measures. Shapley’s solution allocates the importance of each predictor fairly even in the presence of high multicollinearity in the data.