Two control philosophies

The president of a well-known consulting firm does not believe in quantitative objectives for salespeople. He claims that instead of focusing on quantitative objectives by asking a salesperson to meet a given sales volume target, it is better to provide qualitative objectives, such as prospecting a target segment of customers. According to this company president, sales managers should not limit themselves to assigning individual objectives because, although individual objectives induce salespeople to remain committed, only collective action makes the difference.

The president of another major corporation holds quite a different view and asserts that qualitative objectives do not work with salespeople. According to him, only two objectives matter: sales volumes and gross margins. He was reported to say in an executive meeting: “One should stop talking nonsense with those qualitative objectives, number of calls, and so on that lead to evaluate salespeople according to subjective criteria. In addition, selling is an individual sport. Consequently, except for the year end party when the total objective is met, I cannot see what can be done in common.”

How can such dramatically opposed points of view be reconciled? Which of these two executives speaks the truth?

In fact, there is some truth in both arguments. They are only limited by the fact that each executive gives his own vision of the sales force dynamic management process. However, these are far from being incompatible.

The dynamic management and sales force size control at Syntex Laboratories

A few years ago, Syntex Laboratories implemented a decision support system to estimate the optimal size of the sales force they should use (Lodish, Curtis, Ness, and Simpson 1988). Managers at different hierarchical levels estimated the sales market sensitivities to selling effort for various product lines and market segments. The sales force resources (reflected by mentions of a product or details) were optimally allocated to products and market segments. The objective was to allocate salespeople's time to those calls and mentions that were the most profitable to the firm. The final solution was provided as a list of the average number of mentions of each product that should be made to clients in every market segment.

The total time resources (and consequently, the required number of salespeople and the size of the sales force) were gradually increased until net profits started declining. As a result, it was possible to estimate the optimal sales force size and the optimal allocation of the resources simultaneously.

The procedure led to an increase in the size of the sales force from 433 to over 700, which was much larger than had been anticipated by management. The Syntex managers, however, had developed such confidence in the procedure that had been followed during the decision-making process, that they decided to accept those conclusions and dramatically increased their sales force.

Successive sales organization changes at Hewlett-Packard

A few years ago, Hewlett-Packard adopted a completely new sales organization structure. The company was split into several autonomous divisions that drew their resources from headquarters, and that were entrepreneurially and autonomously managed, with complete responsibility for marketing and sales.

In addition, HP established a new CSO division (Computer Systems Organization) which was responsible for marketing a whole line of computers, ranging from individual work stations to powerful units for heavy users. By adopting more open systems, the major selling arguments shifted from equipment's performance and price to customers' added value. The new emphasis was to help customers to find the best possible solutions to their problems, even if this meant using components from a competitor.

In order to provide those services, HP salespeople needed to acquire strong expertise in their clients' problems. HP changed the geographical organization of its sales force so that it was organized by customer type (banking, insurance, industrial buyers, etc.) This was necessary so that the sales staff could understand customer problems within their sector.

However, HP soon realized that this sales force organization was not optimal. Because the salespeople always faced the same type of problems, they also tended to sell the same kind of products and so lost experience in selling the whole line of HP products, from computers to printers.

Why write a new book on sales force management? Obviously, sales and marketing managers have long supervised and controlled salespeople without resorting to the concept of a dynamic sales force management process. Nevertheless, this integrated vision of the control and management of a sales force is likely to add a new dimension to the “classical” conception of sales force management. This new integrative dimension brings together the multiple and complex aspects of a sales force system. This book shows how sales managers can be more effective when they develop such an integrative vision of their mission.

The concept of dynamic management process encompasses more than the traditional managerial functions of situation analysis, planning, program implementation, and control. It includes also the definition of strategic objectives, the translation of those mission objectives into operational and tactical plans and decisions, the constant adjustments of these plans over time, as well as the necessary adaptations of the organization during the process.

The dynamic sales force management process concept relies on a clear vision of a succession of sales missions to be accomplished over time by all the members of a sales team. At each level, the various members of a sales force develop their strategies and tactics that must all contribute to fulfilling the overall missions. The proposed concept builds on the prevalent philosophy of customer relationship management (CRM), and the development of long-term customer relationships that are mutually beneficial.

The field of product innovation has expanded rapidly, and insights regarding the relationship between product innovativeness and new product performance have become clouded, as findings are increasingly mixed. To address this issue and add clarity, the authors quantitatively analyze the extant product innovativeness–new product performance findings. They find that while the resulting relationship between innovativeness and performance is small on average, it lacks generalizability because of a number of measurement (e.g. definition of newness and nature of performance data) and contextual factors (e.g. goods versus services) that moderate the magnitude of the product innovativeness effect found. They subsequently discover that the magnitude of the relationship also has diminished over time as competitive conditions have unarguably intensified. The authors explore the implications of these findings and the revised contingency perspective for academic research and business practice.

As a research base expands, it becomes critical to take stock of extant findings to ensure that previous conclusions and perspectives remain valid and to further ensure that the proper approach to research is being pursued. This assessment becomes even more critical when an area is relevant to practitioners and is attracting a growing number of researchers. Such is the case with the literature focusing on the role of product innovativeness in the marketplace performance of new product offerings.

An extensive empirical literature seeks to analyze the effects of strategic marketing choices on performance using non-experimentally generated, or “real-world,” data. Research utilizing the PIMS database is an important example. Correct treatment of the strategic choices recognizes that they can be “endogenous” for a variety of reasons, including, but not limited to, reverse causation, simultaneity, omitted variables, sample selection, and measurement error. The standard approach for correcting this problem is to use instrumental variable techniques, such as two- and three-stage least squares, in an attempt to purge the endogenous variation from the strategic choice variables, when these variables are used as regressors in a structural equation.

In estimating simultaneous equation models, three important specification issues arise. First, which of the explanatory variables are potentially endogenous? Second, which of the exogenous variables can be excluded from each structural equation? Third, how important are these excluded exogenous variables as predictors of the included endogenous variables? Accompanying these three issues are three specification tests: the endogeneity tests of Wu (1973) and Hausman (1978), Basmann's (1960) test of overidentifying restrictions, and a simple F-test of the explanatory power of the excluded exogenous variables in the “first stage,” or reduced form of the model. Surprisingly, these tests have not been implemented in tandem to any extent in the literature. As a result, many of the results in this literature should be viewed with caution at best, and skepticism at worst.

This volume contains essays that revisit the ideals of the PIMS (Profit Impact of Marketing Strategy) project. They are collected and published here in honor of Robert D. Buzzell's contributions to marketing research in general and the PIMS project in particular. The impetus for these essays originated from a conference held in October 2002. A group of scholars and researchers gathered at the University of Virginia's Darden School to honor Bob Buzzell and exchange ideas and papers reflecting on the achievements and recent advances relating to the PIMS program of research on marketing strategy. What did we learn and what should we have learned from the PIMS project concerning the economic causes and consequences of marketing decisions?

The following people attended the conference:

• Kusum Ailawadi, Tuck School, Dartmouth College

• Jay Bourgeois, The Darden School, University of Virginia

• Eric Boyd, The Darden School, University of Virginia

• Robert Buzzell, Georgetown University

• Markus Christen, INSEAD

• George Day, The Wharton School, University of Pennsylvania

• Paul Farris, The Darden School, University of Virginia

• Bradley Gale, Customer Value, Inc.

• Hubert Gatignon, INSEAD

• Lutz Hildebrandt, Humboldt University, Berlin

• William Kehoe, McIntire School of Commerce, University of Virginia

• Trey Maxham, McIntire School of Commerce, University of Virginia

• Marian Moore, The Darden School, University of Virginia

• Michael Moore, The Darden School, University of Virginia

• Russ Morgan, University of Utah

• Bill Moult, Marketing Science Institute

• Mark Parry, The Darden School, University of Virginia

• Jack Pendray, retired

• David Reibstein, The Wharton School, University of Pennsylvania

• Keith Roberts, PIMS Europe Ltd.

This chapter reviews some key hypotheses from empirical research on success factors in marketing. These hypotheses on drivers of business profitability, in particular quality and market share, have been a major subject of critique, and these critiques have come primarily from the resource-based view in management research. According to this perspective, general laws of business success based on manageable strategic input factors do not exist. Instead, unobservable, firm-specific variables are regarded as the key drivers of profitability. However, only a few studies have been able to show that strong relations discovered in empirical success factor research disappear if unobservable variables are controlled in econometric models.

In this chapter, we show that some of these results may be methodological artifacts. Based on the hypotheses of Phillips, Chang, and Buzzell (1983) regarding the effects of quality and market share on profitability, we use PIMS data to replicate their study using a modified modeling approach. Whereas Phillips, Chang, and Buzzell use data taken at two points in time to investigate the relationships between some key variables, this chapter uses a six-year cross-section of time series and a panel modeling approach to estimate the parameters. This approach allows us to overcome some major objections to the traditional PIMS approach; key relations between observable success factors and profitability highlighted by the PIMS research can be estimated while simultaneously the effects of different types of unobservable firm-specific factors can be controlled.

Over the past twenty years, numerous empirical research studies have examined the magnitude of market pioneer advantages and the associated sources of these advantages. Review articles by Kalyanaram, Robinson, and Urban (1995); Vander Werf and Mahon (1997); Kerin, Varadarajan, and Peterson (1992); Lieberman and Montgomery (1988, 1998); and Robinson, Kalyanaram, and Urban (1994) summarize these research studies.

The PIMS data played an important role in helping to start this research stream and advancing it over time. Because most of this empirical research has been done in marketing, Scherer (1994: 173) says, “we are in debt to business scholars for illuminating the relevant relationships.” The following discussion summarizes the important role PIMS data played in examining order of market entry topics. These topics cover market share, return on investment, firm skills, and marketing strategy development. Future research topics are also discussed.

Order of market entry and market share

In the late 1970s, industry studies in the economics literature documented long-lived market share advantages for first entrants in two specific markets: pharmaceutical products (Bond and Lean 1977) and cigarettes (Whitten 1979). Still, because market pioneers in other markets such as ballpoint pens (Reynolds International Pen), hand-held electronic calculators (Bowmar Instruments), and diet colas (Royal Crown Cola) were quickly overtaken by later entrants, it was not clear whether market pioneer advantages typically surfaced in most North American markets.

The Profit Impact of Marketing Strategy (PIMS) project, which began in 1972, was one of the most successful and influential partnerships between marketing academics and the private sector. Robert Buzzell, as Executive Director of the Marketing Science Institute, was one of a small group of people who made the PIMS project possible. The program resulted in a unique dataset used to investigate the links among marketing strategy, market structure, and performance. The Marketing Science Institute was a near-perfect organizational platform from which to launch a project that had the ambitious goal of understanding how and why some marketing strategies were more profitable than others. To enable this investigation, PIMS, from the beginning, set a new standard of depth and breadth for panel data collected from operating business units. In this book we have collected a set of original essays that revisit the ideals of the PIMS project. Our purpose is to explore what we learned and, perhaps, what we should or still might learn about researching the connections between marketing strategy and profits.

This does not mean that we are finished with the questions that PIMS helped the field of marketing strategy pose. However, enough time has passed and enough additional evidence has been accumulated that we believe it is appropriate to appraise what was accomplished. Some of the essays will help put the achievement of PIMS into the context of the times (both then and now).

For researchers investigating questions related to marketing strategy and financial performance, there are few databases that are, in any way, comparable to PIMS. However, COMPUSTAT is one database that researchers have used frequently to address such questions. Other databases that provide limited additional points of comparison are that of the Inland Revenue Service (IRS) and the Federal Trade Commission (FTC) Line of Business data. In this chapter we will provide some comparisons between PIMS and COMPUSTAT data.

From the beginning, researchers were impressed by the “overwhelming superiority of PIMS data to other sources in quantity, number of measured variables, timeliness, [and the] conscientious attempt to minimize potential sources of input error” (Anderson and Paine 1978). Of course, the “timeliness” of the PIMS data is no longer a strong point and, since 1990 or so, publications of empirical findings based on the PIMS data have appeared far less frequently in major marketing and strategy journals – published articles have declined along with the size and currency of the data.

On the other hand, publications based on COMPUSTAT data are appearing with increasing frequency in marketing journals. Further, many of the issues addressed by researchers using COMPUSTAT are similar to those addressed with analyses of the PIMS data. For example, both COMPUSTAT and PIMS data have been used to study: relationships of market share, firm size, and power to profits; determinants of marketing cost ratios and media budgets; and returns from R&D and new products activities, and patents.

More than twenty years ago Farris and Reibstein (1979) published research that demonstrated a strong cross-sectional correlation between relative advertising expenditures and relative prices charged by manufacturers of non-durable consumer goods. Data for that research were taken from the PIMS database. The correlation was demonstrated to survive a number of controls for relative quality and market share. The correlation was also shown to be stronger for later stages in the product life-cycle and for products purchased in relatively small dollar amounts. The research made no claims about the direction of causality from advertising to prices or vice versa. Instead, the paper argued that from the management perspective “consistency” between advertising and pricing was important. In other words, businesses with high (or low) relative prices should generally also have high (or low) levels of relative advertising. The claim for the importance of consistency was buttressed by evidence in the paper that businesses with inconsistent pricing and advertising strategies earned lower ROIs.

In this chapter we first review and then extend the earlier Farris and Reibstein (1979) study with new analyses based on the PIMS data. The review is placed in the context of a broader managerial (not necessarily a public policy) concern with the relationship between total marketing costs (not just advertising) and prices. The expanded view of marketing costs includes salesforce and other marketing expenses – budget items with collective dollar values that are typically three to four times advertising budgets.

Simple econometric models often produce results that may be interpreted in different ways. In response to disagreements over how to interpret such models, researchers have begun to apply increasingly sophisticated econometric models and estimation techniques. However, it is not always clear that available data are appropriate for the task presented by the more sophisticated models. In this chapter we address such a problem.

The controversy

One of the key early findings from the PIMS database was the positive effect of market share on business profitability (Buzzell and Gale 1987; Buzzell, Gale, and Sultan 1975). The argument was that higher market share yielded advantages in efficiency and thus resulted in lower average cost. Based on the empirical evidence from PIMS and other studies, such as those with the Federal Trade Commission's Line of Business database (e.g. Ravenscraft 1983), Scherer et al. (1987) contended that the market share effect is “robust,” and a number of analysts and consultants promoted the unbeatable logic of market share building strategies (e.g. Henderson 1979).

However, a number of researchers raised serious questions about the validity of the market share effect. Some pointed out that the observed regularity lacked a theoretical base (e.g. Rumelt and Wensley 1980). In fact, knowledge of such a strategic relationship would force all firms to compete more forcefully for market share, which would eliminate the returns implied by the relationship, unless “isolating mechanisms” existed that limited competition (Wensley 1982).