About AMA

Executive Summary
Given an unlimited marketing budget, managers can contact all their customers in every business cycle. Such a strategy minimizes the Type I error of not contacting a customer who could have potentially provided revenue. However, minimizing Type I error also maximizes a so-called Type II error, or contacting a customer who is not ready to purchase and is costly in terms of adversely affecting both the bottom line and the top line. When firms are faced with a limited marketing budget, the trade-offs between Type I and Type II errors are highlighted further, and managers are forced to prioritize their communication strategies toward customers who are expected to provide the highest growth in cash flows (i.e., customer selection). Thus, accurate measurement or estimation of the future value of customers or customer lifetime value (CLV) is critical for the success of customer selection in a firm. Although CLV is considered a theoretically superior metric for customer selection, widespread adoption of CLV among firms is limited because of two practical challenges.

First, given the uncertainty inherent in predicting customer behavior, managers are wary about taking actions based on CLV, which is calculated using predictions of customer behavior over the long run. Typical sources for this uncertainty are related to the poor or nonexistent information in most customer relationship management databases on (1) customer transactions with the competition, (2) competitor marketing actions targeted at each customer, and (3) customer attitudes. Sometimes, the cost of increased errors can outweigh the benefits of long-term predictions. Second, there is no guidance on a forward-looking customer-level cost allocation rule when computing CLV. The practice of using status quo costs as an estimate of future costs of serving a customer is inconsistent with the core assumption for using forward-looking metrics over backward-looking metrics (i.e., previous customer profitability is not the best indicator of future customer profitability). Thus, similar to the metrics, the cost allocation rules also need to be forward looking.

To address the preceding practical challenges in using CLV, the authors (1) provide a Bayesian decision theory–based customer selection framework that accommodates for the uncertainty inherent in predicting customer behavior, (2) propose a joint model for predicting customer purchase timing and purchase quantity that is amenable for a Bayesian decision theory–based customer selection process, (3) compare the proposed customer selection framework with a collaborating firm's current customer selection framework, and (4) evaluate the total profit implications of various customer-level cost allocation rules. The collaborating firm in this study is a large multinational firm that sells high-technology products and services in the business-to-business market. The managers face the decision of determining the level of resources to allocate in each communication channel to maximize the expected CLV. This decision must be made using predictions about future customer behavior, such as purchase timing and quantity.

On the basis of the analyses, the authors infer that (1) accounting for the dependence between purchase timing and quantity leads to a more accurate estimation of CLV, (2) selecting customers using a Bayesian decision theory–based customer selection framework leads to a better identification of profitable customers, and (3) the optimal costs derived based on estimates of customer's historic responsiveness to marketing communication provides a good estimate of the expected costs of serving a customer and aids in identifying profitable customers.

The authors provide several guidelines for implementation that can be broadly classified as follows: (1) plan for a longer period for model estimation, (2) align model reestimation cycles with resource allocation cycles, (3) use an optimal level of marketing contacts as a benchmark, (4) incorporate salesperson intuitions, and (5) evaluate model reformulation. On the basis of the results from the analyses, the authors provide recommendations to mangers who want to enhance marketing productivity and estimate returns from marketing actions.

Biography
Rajkumar Venkatesan has been on the faculty at the University of Connecticut and received his PhD from the University of Houston. Raj's research focus is on designing marketing strategies that maximize customer profitability, understanding the pricing strategies of online retailers, and developing models for forecasting sales of new products. His research has appeared in several journals, including Journal of Marketing, Journal of Marketing Research, Marketing Science, and Harvard Business Review. Raj's research has been recognized with awards such as the Donald R. Lehmann Award for the best dissertation-based article published in Journal of Marketing and Journal of Marketing Research,the Marketing Science Institute Alden G. Clayton Award for the best marketing dissertation proposal, and the ISBM outstanding dissertation proposal award.

V. Kumar has been recognized with more than 15 teaching and research excellence awards including the Paul H Root Award (two times) for the article published in Journal of Marketing that best contributes to the practice of marketing, and the Donald R. Lehmann Award (two times) for the best article published in the Journal of Marketing or Journal of Marketing Research over a two-year period. Recently, one of his articles in forecasting won the Outstanding Paper Award from the International Institute of Forecasters. He has published more than 75 articles in many scholarly journals in marketing, including the Harvard Business Review, Journal of Marketing, Journal of Marketing Research, Marketing Science, and Operations Research. He has coauthored multiple textbooks on marketing research. He has authored a book titled International Marketing Research, which is based on his marketing research experience across the globe, and his book Customer Relationship Management: A Databased Approach was recently released. His current research focuses on multichannel shopping behavior, international diffusion models, customer relationship management, customer lifetime value analysis, sales and market share forecasting, international marketing research and strategy, coupon promotions, and market orientation. He has taught in several universities and organizations worldwide. He was recently listed as one of the top-five ranked scholars in marketing worldwide. He has consulted for many global Fortune 500 firms. He received his PhD from the University of Texas at Austin.

Timothy Bohling has held several marketing leadership positions in IBM. Currently, as director of IBM Global Business Services Market Intelligence, Tim is responsible for the formulation and execution of the market intelligence plan, which delivers thought leadership, deep insights, and recommendations that drive IBM to take better action, improve performance, and fuel growth. Tim directs the Worldwide Market Intelligence program, which consists of market and customer analysis, primary research, secondary research, opportunity analysis, database marketing, database analytics, and competitive intelligence. Before joining IBM, Tim held senior marketing management positions with GTE and Fogarty Klein & Partners. Tim's consulting experience includes many Fortune 50 companies including AT&T, Exxon, Texaco, PepsiCo, and Bank of America, among others. Tim is an accomplished public speaker and frequently presents at industry conferences and seminars. His research has been published in several industry journals, including Marketing Science, Journal of Industrial Marketing Management, and Journal of Interactive Marketing, among others. Tim received a BBA in Marketing and an MBA with highest honors from the University of Houston.