Customer Lifetime Value Improved Based On Online Review Behavior

Customer lifetime value has always been the most important support for enterprises to manage the customers and do the relative marketing activities. By calculating customer lifetime value in both the aggregate level and individual level, enterprises can recognize the customers’values so that they can adjust the customer management direction to realize the maximum of customer value.This paper focused on the two dimensionalities:the aggregate level and the individual level, by using two different mathematical models to calculate the customer value so that we can lay the foundation for the marketing decisions.During the era of web2.0, more and more customers are willing to make comments about the relative products, services, brands and comments about the enterprises on the purchase, review and SNS websites. As we all known, for the enterprises, each customer can bring direct value by purchase behavior. But we don’t know whether the review behavior can also bring values for the enterprises or not.Nowadays, researchers only analyze the customers’purchase behaviors and make the mathematical model from this dimensionality, ignoring the review behaviors and do not analyze the purchase and review behaviors together. Because the view is relative new and lack of studies, we decide to rebuild the mathematical model to calculate customer lifetime value by considering the purchase and review behaviors together.While studying the Dianping.com, we found that one customer can both use the purchase platform and review platform so that these two behaviors can influence each other. Because of this characteristic, we try to build the aggregate level and individual level models to recalculate the customer lifetime value.The models of our paper are composed by two parts:the aggregate level and the individual level. For the aggregate level, we use the RFM model as the basic model, and bring the review behavior into the basic model, proposing the improved RFM model considering two behaviors; For the individual level, we first use Logit model to analyze the influence between two behaviors and get the important variables. Then put these important variables into the Pareto/NBD model as covariant. Next, we use the traditional Gamma-Gamma model to predict the purchase monetary. Meanwhile, we also use the Logit and Pareto/NBD to predict the review times. Finally, we build the customer lifetime value mathematical model. The result find that the prediction model have good effect.Our research finds that the customer’ s purchase behavior can influence their own purchase behavior by using the real data from Dianping.com. This result can support the enterprise to recognize high value customer and provide strong support of the relative management measures.