Recent research has empirically characterized the buyer.seller relationship as dynamically evolving from one discrete state to another. Conventional wisdom would suggest that a customer in a higher relationship state that has a higher transaction value would also have greater lifetime value to the firm. However, recent evidence suggests that higher relationship states can be ephemeral. Hence, the link between transaction value and lifetime value is not obvious. In this study, we seek to understand, within a specific empirical context, (i) the relationship between a customer's transaction value and that customer's lifetime value and (ii) the relationship between the lifetime value of a customer and the optimal level of marketing activity that needs to be directed at that customer. To this end, we develop a trivariate Tobit hidden Markov model that allows for (a) transitions among relationship states, (b) possible synergies between the various products that the supplier firm offers, (c) endogeneity in marketing activity, (d) heterogeneity in model parameters, and (e) the presence of the nopurchase option. Our results reinforce recent findings by Schweidel et al. [Schweidel, D. A., E. T. Bradlow, P. S. Fader. 2011. Portfolio dynamics for customers of a multiservice provider. Management Sci. 57(3) 471.486] that higher relationship states can be short-lived. Importantly for the supplier firm, a customer in the highest relationship state in a given period does not yield the highest lifetime value to the firm. Hence, the relationship between transaction value (i.e., relationship state) and lifetime value can be nonmonotonic. At the same time, we also find a nonmonotonic relationship between the optimal expenditures that should be directed at a customer and that customer's lifetime value; i.e., the optimal level of marketing contacts is not the highest for customers with the highest lifetime value. Furthermore, we find that the optimal marketing expenditures for myopic agents are 14%.33% lower than the corresponding values for forward-looking agents. Therefore, not accounting for the long-term effects of marketing contacts would lead to suboptimal marketing budgets. Moreover, a comparison with the current marketing expenditures suggests that the current practice is closer to the myopic policy than to the forward-looking one.