Dynamic Marketing Resource Allocation with Two-Stage Decisions

In the precision marketing of a new product, it is challenge to allocate limited resources target customer groups with different characteristics. We presented framework using distance-based algorithm, K-nearest neighbors, and support vector machine capture customers’ preferences toward promotion channels. Additionally, online learning programming was combined strategies fit dynamic environment, evaluating its performance through parsimonious model minimum regret. A resource optimization proposed classification results as input. particular, we collected data from an institution that provides financial credit products capital-constrained small businesses. Our sample contained 525,919 customers who will be introduced product. By simulating scenarios between demand, showed up 22.42% increase in number expected borrowers when KNN performed optimal allocation strategy. also show most stable method perform algorithm has efficient adoption learning.