Footnote:
Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments January 16, 2022 erstellt
Description:
One important problem in the marketing and user growth for businesses is to infer the marginal gains of specific campaigns or strategies, which is crucial for the decision makers to determine whether to carry on or terminate certain policies. However, in most cases the return of gaining customers is affected by the cost in an asynchronous way, which leads to the difficulty in assessing or optimizing the policies. This paper proposes a predictive model, which stacks the neural network and random forest, to predict the Remnant Lifetime Value (RLV) of individual users, that is a generalization of the Customer Lifetime Value (CLV). Based on the predictive model, a heuristic framework called Dynamic Profit-based Allocation-Execution Framework (DPAEF) is introduced to solve the dynamic allocation-execution problem. Both the predictive model and the framework are shown superior to the typical baseline models via conducting corresponding A/B tests. We also discuss the generalizations of this method to some circumstances where the extra limitations or tools exist