From the perspective of enterprises, the paper effectively unifies Hidden Markov model and the dynamic programming technology to analyze quantification, decisionmaking as well as optimization of single person's CLV. It proposes the dynamic model of individual customer's CLV and tests the model. The model could be used for parametric estimation using the observed customer purchase decision sequence as the solution of the dynamic optimization problem. The model could also be used for dynamic estimation and classification of customers according to the sequence, and judgment of the relation phase of firm and customers. Thus it provides a quantitative method of dynamic customer relations management for enterprises and lays the foundation of firm for the implementation of the optimal marketing strategy.