客户关系动态优化模型与实证研究

提出了客户关系与营销活动的动态交互模型,以长期收益最大化为目标,优化企业的营销活动。模型假设客户关系可离散为几个层级状态,并设客户关系所处状态受营销活动的影响而动态的变化,服从马尔可夫决策过程。客户关系状态所处层级不可直接观测,但其与客户购买水平有概率相关关系。提出模型参数估计的最大似然估计方法。以国内某企业的客户关系管理数据为例,说明了模型变量的定义方法,通过客户交互历史数据估计模型参数,并对客户管理策略进行优化。结果表明,最优策略管理下期望提升客户价值61％～82％。

Customer Relationship Dynamic Optimization Model and Empirical Study

In this paper, we propose a dynamic model to capture the interactivity between marketing actions and customer relationship. The model uses long term rewards as the objective to optimize marketing actions. We assume that customer relationship can be discretized to several levels, and the relationship levels customer stays in is affected by marketing actions. This dynamic process is assumed as a Markov process. Which levels customer stays in is not observable directly, but it is probabilistically correlated with customer purchasing levels. To obtain the model parameters, we further propose a maximum likelihood estimation approach. In empirical study, we apply the model to a dataset from a company in China. We show how to define the model variables, how to estimate the model parameters, and finally how to optimize the customer relationship management policies. The results show that we can increase expected customer lifetime value by 61％～82％ when using the optimized customer management policy.