Procter & Gamble (P&G) Asia-Pacific is interested in managing value growth. Only after fully understanding the true effects of the marketing-mix variables can P&G managers make strategic decisions answering questions such as the following: (1) Are the P&G brands in the detergent market inelastic or elastic with respect to price? How has the price elasticity changed over time? Can P&G increase the price of its brands to gain value growth? (2) What are the price, distribution, and sizing combinations needed to achieve the desirable value growth? (3) How can P&G gain market share from its competitors without cannibalizing its own brands? P&G Asia-Pacific approached us to develop a value growth framework to answer the above questions. To generate the answers for the above questions, we develop a three-step weighted random coefficient estimator that captures the heterogeneity across cross sections (different stock-keeping units and states) and the endogeneity of distribution. Based on the parameter estimates, we provide strategic recommendations to P&G for a field test to validate our suggestions. We developed a simulator for P&G managers so that they can generate appropriate marketing-mix strategies for achieving the desired value growth. As a result, P&G gained over $39 million in value growth over a one-year period by implementing the recommendations from our modeling approach.

Customer management activities at firms involve making consistent decisions over time, about: (a) which customers to select for targeting, (b) determining the level of resources to be allocated to the selected customers, and (c) selecting customers to be nurtured to increase future profitability. Measurement of customer profitability and a deep understanding of the link between firm actions and customer profitability are critical for ensuring the success of the above decisions. We present the case study of how IBM used customer lifetime value (CLV) as an indicator of customer profitability and allocated marketing resources based on CLV. CLV was used as a criterion for determining the level of marketing contacts through direct mail, telesales, e-mail, and catalogs for each customer. In a pilot study implemented for about 35,000 customers, this approach led to reallocation of resources for about 14% of the customers as compared to the allocation rules used previously (which were based on past spending history). The CLV-based resource reallocation led to an increase in revenue of about $20 million (a tenfold increase) without any changes in the level of marketing investment. Overall, the successful implementation of the CLV-based approach resulted in increased productivity from marketing investments. We also discuss the organizational and implementation challenges that surrounded the adoption of CLV in this firm.

In this paper, we provide theoretical arguments and empirical evidence for how Genetic Algorithms (GA) can be used for efficient estimation of macro-level diffusion models. Using simulations we find that GA and Sequential Search-Based-Nonlinear Least Squares (SSB-NLS) provide comparable parameter estimates when the data including peak sales are being used, for a range of error variances, and true parameter values commonly encountered in the literature. From empirical analyses we find that the forecasting performance of the GA estimates is better than that of SSB-NLS, Augmented Filter, Hierarchical Bayes, and Kalman Filter when only pre-peak sales data is available for estimation. When sales data until the peak time period are available for estimation, SSB-NLS is able to obtain parameter estimates when the starting values provided are the estimates from using GA. The estimates from GA are not biased and do not change in a systematic fashion when postpeak sales data are used, whereas the estimates from SSB-NLS are biased and change in a systematic fashion. Summarizing, we find that GA may be better suited for diffusion model estimation under the three conditions where SSB-NLS has been found to have problems.

The literature on cross national diffusion models is gaining increased importance today due to the needs of present day managers. New product sales growth in a given nation or society is affected by many factors (Rogers 1995), and of these, sociocontagion (word of mouth) has been found to be the most important factor that characterizes the diffusion process (Bass 1969, Moore 1995). Hence, it is interesting and perhaps challenging to analyze what would happen if a new product diffuses in parallel in two neighboring but culturally different countries. Not only will we expect the diffusion process in the two countries to be different, but we will also expect some interaction among them, especially if the two societies mingle with each other. There are two streams of research in cross national diffusion. The first type focuses on exploring the differences between diffusion processes in two countries and finding out whether those differences can be attributed to social and cultural difference between the countries involved. Examples of this type of research are found in Takada and Jain (1991), Gatigon et al. (1989), Helsen et al. (1993), and Kumar et al. (1998). These studies did find some relationship between the cultural differences of the countries studies and the differences in the diffusion process. The second stream of research focuses on modeling explicitly the interaction between the diffusion processes in two countries. The interaction is typically captured through lead-lag effect (Eliashberg and Helsen 1996, Kalish et al. 1995), where the sales process in the lead country (i.e., the country where the product was first introduced) is modeled to affect the sales process in the lag country (i.e., the country where the product was introduced a few years later. Another method to study the interaction among the diffusion processes in two countries was suggested by Putsis et al. (1997), who used a "mixing model" to empirically explore the existence of such interactions. These studies basically observed that, when a new product is introduced early in one country and with a time lag in subsequent countries, the consumers in the lag countries learn about the product from the lead country adopters, resulting in a faster diffusion rate in the lag countries. Ganesh and Kumar (1996) formulized this effect as the learning effect to be influenced by country-specific factors (cultural similarity, economic similarity, and time lag elapsed between the lead and the lag countries) and product specific factors (continuous vs. discontinuous innovation and the presence or absence of a standardized technology). A careful analysis of the extant literature on the second stream of research would reveal that neither the learning effect model nor the mixing model can be modified to accommodate the other model. Our contribution to the literature exactly addresses this point. In this paper, an alternative framework is proposed that has two unique features. First the framework is flexible enough to not only account for the lead country affecting the lag countries and vice versa, but also to accommodate the simultaneous interaction among countries in explaining the diffusion processes in the countries concerned. Using multiple product categories and a variety of new product introduction situations, we empirically demonstrate the flexibility and efficiency of our proposed framework. We found strong evidence of all types of interaction, namely, lead lag, lag lead, and simultaneous, which evidence suggests that one cannot afford to omit any of the interactions. The second unique feature of our paper is the estimation procedure that we used. Because statistical estimation of a dynamic process that includes lead-lag, lag-lead, and simultaneous types of causality within a single framework is not straightforward, we suggest and iterative estimation procedure for the estimation. This new procedure not only proved to be flexible in accommodating difference types of interaction, but also converged rather quickly in all of the cases that we empirically tested. Noting that the statistical properties of these estimators are not generally available, we carried out a simulation exercise that clearly revealed the efficiency of the proposed estimation procedure. After analyzing the interaction, we went further and showed that the magnitude of the cross-national influences is affected by certain country-specific and product specific factors. The flexibility of the proposed method over the existing methods is demonstrated through obtaining superior forecasts with the proposed method. Several interesting insights for managers concerned with formulating international marketing strategies are offered.