Conceptual Foundations

Professor John M. McCann
Fuqua School of Business
Duke University

This section describes a new and different philosophy for building marketing information and decision support systems. The motivation for attacking marketing systems from a new perspective was provided a decade ago by Little:

Managers do not like terminals. They are impatient and busy. They do not formulate problems in model terms because that is not the way they actually think. They want to think about strategy, not analysis. They will propose actions to be analyzed, but they will not do it themselves.[1]

Although great strides have been made by the developers of marketing systems during this decade, these efforts have resulted in systems which aid an analysis, when the manager wants to perform the analysis. Little's view indicates that managers do not, in fact, want to perform the analysis. Rather, they may be willing to indicate which analyses they would like performed, but they would prefer that some other person (or "thing") do the actual analyses and draw the important conclusions for them.

More recent evidence indicates that this situation has not changed since Little's quote. Based upon interviews with marketing managers in 13 consumer products firms, McCann [2] concludes that managers are using computerized information systems for mundane tasks. Goldstein and Zack [3 p. 18] confirm this conclusion. However, existing marketing information systems require that the user do the analysis. But Little's experience indicates that marketing managers want to think about marketing strategy, not numbers and computers. This chapter reports on a different type of research, one aimed at allowing the manager to indicate which results s/he wants from the system. The manager indicates what type of statements s/he wants the computer to make about a particular brand or market, and the computer uses its knowledge to make the statements and present the supporting evidence.

Such a computer system would be very different from existing marketing information systems, which require the manager to think in terms of data, models, and analysis. Such a manager must deal with measures (CDI, Opportunity index, share, trend, etc.) and systems (Express, Metaphor, etc.). These measures and systems are a means to an end: marketing analysis and planning. The goal of the new approach to marketing systems is to liberate marketing managers from the tyranny of data and models by allowing him/her to focus on marketing issues. Such systems incorporate the appropriate data and models to allow the manager to think about the marketing mix elements.

The new systems will allow managers access to a new marketing world, one in which they would prefer to operate. This new world allows them to do what they like to do, design and execute marketing programs, rather than use a computer to answer questions. The term Marketing Gate is used to denote a computer system that opens this world to the manager.

By passing over the Marketing Gate the manager leaves the world of data and equations and enters the world of the marketing mix elements: advertising, promotion, and price. This transition allows the marketing manager to move from being an analyst to being a marketer. "Let marketers do marketing" denotes the fact that the system should allow marketing managers to focus on the marketing tasks rather than the analytical tasks.

Another desirable feature of marketing systems involves a partnership between the manager and the system; the system serves as an advisor to the manager. Imagine two people working together, an Assistant Product Manager (APM) and a Product Manager (PM). The APM is new on the job while the PM has several years experience in product management. The APM is using an MMIS to perform an analysis, and the PM is offering advice throughout the process:

APM: Sales are not performing as expected. Let's see if the problem is localized. Look at that ... the problem is in Washington.
PM: Keep on this track by seeing if it is chain specific or a general Washington problem.
APM: Aha! Giant Food is the problem. We are just not getting good trade support. Let's plan something to solve this problem.
PM: You are probably right, but I think we ought to check out Giant's trade support before we jump into planning. Let's do a promotion review on Giant.
APM: I was right ... the results show that we have gotten no price cuts, and only token feature and display support from Giant.
PM: Yes. No one will argue with us because we have done enough analysis to build a strong case. Now it's time to plan.
APM: What type of marketing program do you recommend?

PM: A trade deal in Washington designed for Giant.

Since the Product Manager is offering advice to the less experienced Assistant Product Manager, he could be termed an advisor or consultant to the computer user, the APM. Marketing Gate plays a similar role, and it could be described in several ways: as an assistant, an agent, an advisor, or even a soft robot. Just as the PM knew what the APM was doing and what results were being obtained from the MMIS, Marketing Gate also knows something about what the manager is doing because it too knows the results.

The Computer's Role in the Analysis Process

Most marketing managers use computers to do analysis, i.e., to gain an understanding of their business and the forces which are having impact on the brand. The manager then summarizes the analysis in a report or a presentation which describes the problems and/or opportunities facing the brand. After the report is completed, the computer is "turned off" and the manager goes on to design and execute programs aimed at solving the problems and/or capturing the opportunities. After the programs have been executed, the manager may turn the computer on again to perform additional analyses to determine the success or failure of the programs.

The usual interaction between the manager and the computer is a type of conversation. The manager has some goal in mind when starting the analysis, but the computer is not informed of the goal. The manager devises an analysis plan which will allow the goal to be achieved and issues commands to the computer during each step of the analysis plan. The computer responds to each command by retrieving data, performing calculations, or reporting results. The manager reads and interprets the results, which may lead him/her to draw some conclusions. The computer is not informed of these conclusions because there is no reason to tell it what the manager has just concluded. The computer provides numbers upon command; it does not have a reason to "know" how the numbers are used; this is not its role. The manager's job is to make inferences and deductions based upon the numbers and to use these conclusions for other activities. The computer is not a partner or participant in these intellectual activities.

A good model of this work involves the manager thinking and the computer calculating. The manager's thoughts are private in the sense that they are not shared with the computer. These thoughts become part of a presentation, which provides the record of the analysis process. Such a record might be useful later when steps have been taken to act upon the results of the analysis. Considerable effort has been expended by the vendors of marketing management information systems to make their systems easy to use, to make it easy for a marketing manager to interact with the system, to make it easy for the manager to inform the computer of the desired tasks and to understand the output. These efforts are aimed at improving the analysis aspects of marketing.

There is much more to marketing than using a computer to analyze numbers. The marketing managers manage the brand, and this management process involves the four managerial tasks of analysis, planning, execution, and control. Before we examine each of these tasks, it might be insightful to examine the practices of a related professional, a medical doctor. A doctor performs a diagnosis on a patient. Based upon this diagnosis, a treatment program is ordered. The treatment is administered and the patient is monitored to determine if the treatment worked and if the symptoms went away. The medical profession has devised a simple but very useful device for communicating this process: the medical chart. The doctor enters the symptoms and diagnosis into the chart, along with the prescribed treatment. Subsequent patient measurements are entered into the chart. Interpretation of the original symptoms and subsequent measurements allow the appropriate medical personnel to understand the state and history of the patient, as well as ascertain the effects of the treatment. This chart is the record of what the doctor knows about the patient, at least those facts and subjective evaluations that the doctor thinks are important.

A significant aspect of this chart is the fact that other medical professionals can benefit from the doctor's original diagnosis. This memory is important because the doctor is usually not the only care provider for the patient. We can think of the chart as "knowing what the doctors know about the patient".

What does all of this have to do with the use of computers by marketing managers? Why is this medical example important for our story? It is important because it allows us to illustrate a shortcoming of almost all marketing information systems: they do not have a memory. People use computers to gain insight, to learn. But this learning is not transmitted to the computer like doctors transmit their findings to the medical chart.

However, a new generation of computer systems is coming onto the scene which goes beyond the traditional systems. Artificial intelligence systems are being built which capture more of the human intelligence processes than do most conventional systems. Builders of AI systems study human intelligence and then build systems which perform some aspect of intelligent behavior. One recent AI book [4] uses the medical doctor example to illustrate the role of AI. After going through a simple example of a doctor diagnosing and treating a patient with chicken pox, the author makes the following observation:

... human beings don't just solve problems and then forget about them. Rather we learn from experiences. ... If the patient in our example were definitely suffering from chicken pox, then this information ought to be stored for future reference ... someone else looking at this person's medical record would not have to rediscover the diagnosis for him- or herself from the original symptoms.[5]

The doctor lets the chart know what s/he knows. The marketing manager should let the computer know what s/he knows? What would be the implication if a marketing system were designed around the phrase "the computer knows what the manager knows about the situation"?

The Marketing Gate system, an initial attempt to build a marketing-oriented system, is based upon the twin themes:

The next section lays the foundation for the application of these themes by exploring the ways a marketing manager uses an information system. The result of this exploration is an initial understanding of new ways to use the computer and a beginning of a science of Era I computing.

Towards a Science of Era I Computing

Marketing managers use computers to gain an understanding of their markets and products. These computer systems are termed Marketing Management Information Systems (MMIS) because they are used by marketing managers to gain access to marketing information. McCann and Gallagher [6] describe these types of marketing information systems as Era I systems, thus placing them at the beginning of an evolution of marketing systems through three additional eras.

The popular systems include Metaphor (currently renamed the Data Interpretation System), Express, Acustar, Scan*Fact PC, Decision Master, Analect, and IN*FACT. McCann [7] describes how marketing managers use such an MMIS. The practices of one firm, termed Beta in the book to disguise the firm, aptly describes the practices of typical marketing managers. When Acustar, the firm's MMIS, was first installed, the marketing managers used it to extract different views of the data. To obtain a view, a manager had to issue a series of commands to the MMIS. This practice led to some standard ways of viewing the data, i.e., to standard reports. Then the manager prepared an Acu-Exec for automating this view. An Acu-Exec stores most of the necessary commands in a file and the report can be run by typing the name of the file. A library of Acu-Execs was available to all marketing managers. These managers performed most of their work by selecting the appropriate Acu-Exec.

This is a standard process for most managers; only the system names are different. Users of the Metaphor system call their procedures "capsules", while users of Express employ the term "reports". An example based on the Express system will serve to illustrate this concept. The following is an example that shows the values of marketing share and retail advertising share for various brands.

              TOTAL UNITED STATES                
         FOR ISSUE 203 ENDING 04.02082           

                     Majers      SAMI               
                     Share      Share      Index    

MAMA MIA'S             9.0       41.5          22    
ITALIA                 7.2        8.3          87    
MARCO                  7.0        4.5         155    
NATURAL                7.2        4.8         149    
PAPA                   5.0        3.0         164    
DELICIOSO              7.4        3.9         192    
OLD WORLD              7.2        2.5         285    
MEDITERRANEAN          7.1        1.7         425    
NAPOLI                 7.0        0.9         750    
PRIVATE LABEL          3.0       15.4          19    
CAPRI                  7.3        5.8         127    

Source: Management Decision Systems [8]

The marketing manager who ran this report could make the following observations:

S/he could then go further and draw a conclusion:

Discussions with about 50 marketing managers over the past six years has led to the realization that they use an MMIS program to make observations and draw conclusions about brands, customers, consumers, and/or competitors. Marketing analysis is performed by most marketing managers by the repeated application of the process of (1) selecting a report-type to be run, (2) running the report on the appropriate data, (3) making observations from the data, and (4) drawing conclusions from the data and/or the observations. In a sense, this is the "science of Era I computing".

The Problems

The above procedure works for most marketing managers; their MMIS coupled with their cognitive and motor skills allow them to run the appropriate reports and make the appropriate interpretations. The world of consumer marketing, however, is changing rapidly.

The microprocessor is bringing about a number of technological developments which effect the data available to marketing managers. Two important results are (1) an explosion in the quantity and quality of marketing data and (2) the automation of the retailing process. These two factors are changing the practice of marketing by shifting the focus to local markets and retail accounts, and making real time marketing possible.

All of these changes are resulting in manager overload. Marketing managers in charge of the brands have more opportunities than they do time and other resources to capture them. Overload results in missed opportunities. Thus it is necessary to enhance the practice of marketing analysis, planning, execution, and control. Overload also demands changes in marketing systems to improve the efficiency and effectiveness of marketing, and to enable marketing managers to do tasks they cannot currently perform.

The problems involve time, especially the amount of time available to perform the tasks which senior management expects from marketing managers. The following comments were made by a division General Manager in a large consumer packaged goods firm:

This firm runs on auto-pilot. We have flexible manufacturing systems and use a just-in-time philosophy, which allows us to turn on a dime. Our sales force is well disciplined and very effective. Our promotion department is staffed with professionals who know what they are doing. The brand teams seem to be the bottleneck. I bought them a Metaphor system and we have seen what can happen when they get going on it. But, they just seem to scratch the surface. We get a few good marketing programs out of them. We could do so much more.

Brand managers respond with statements like the following:

There are so many markets and so much data; we have so many opportunities. We know how to do the work; we simply do not have the time.

This mismatch between expectations and performance is causing marketing managers to spend time on activities that are tactical and analytical in nature rather than on the more strategic aspects of the business. A study [9] of how product managers allocate their time indicates that over 40% of their time is spent on promotion, forecasting & budgeting, and analysis. And, these managers would like to decrease the time spent on these activities and increase the time spent on advertising and general marketing.

The planning aspects of marketing management can be quite time consuming, both in terms of the amount of time spent during the formal planning process and in the ad hoc analysis and planning which occur after the planning cycle is completed. For instance, the marketing plan development process at Procter & Gamble [10] involves the following tasks over a a 12 week period: business review, competitive forecast, preliminary forecast, promotion review, media plan, and budget proposal. An even more time consuming planning process was practiced at Pepsi [11], where the strategic and marketing planning process started in mid-January and did not end until late May.

These planning processes are designed to bring the firm's marketing and brand knowledge to bear on the development of future marketing programs. For instance, a promotion review is performed during the P&G planning process which calls for the brand group to review results of all promotions on all brand items in all markets. The goal is to explain the reasons for success and failure. To perform this review, the brand group must know how to do a promotion review and how to interpret the results. Most of this type of analysis is actually performed by the newest member of a brand group, typically called an Assistant Product Manager (APM). This APM must acquire the necessary knowledge about the firm's accepted methods for doing a promotion review, and apply this knowledge to every item in every market for every promotion event. If a firm has 25 brands, and each brand promotes 3 different items an average of 4 times in 25 markets, the result is 7,500 events which must be analyzed. Then a meta-analysis [12] could be performed to explain the reasons for success and failure in each of the 25 brand groups.

After this analysis is completed, the brand groups must design promotions for the following year. In our hypothetical firm, the result would be the design of 7,500 promotions. Whereas the promotion review required analytical knowledge, the design of promotions requires marketing knowledge, along with institutional knowledge about the promotion processes involved in promotion fulfillment houses, media, and the retail trade.

These analytical and marketing planning activities are an exercise in applying knowledge to marketing, which we define as understanding and influencing markets. The promotion review is an example of the "understanding" aspect of marketing, and the promotion planning is an example of the "influencing" aspect of marketing. Marketing Gate is an initial attempt at computerizing analytical and marketing knowledge to improve the efficiency, effectiveness, and enabling capabilities of marketing systems. The next chapter describes the notion of a report writing assistant as the focal point for this new class of software application.


  1. John D.C. Little, "Decision Support Systems for Marketing Managers", Journal of Marketing, Vol 43, Summer, p. 22, 1979.
  2. John M. McCann,, The Marketing Workbench, Dow Jones-Irwin, 1986.
  3. David K. Goldstein and Michael H. Zack, "The Impact of Marketing Information Supply on Product Managers: An Organizational Information Processing Perspective", Working Paper 88-054, Harvard University, Graduate School of Business Administration, March, 1988.
  4. Nigel Ford, How Machines Think: A General Introduction to Artificial Intelligence, Wiley, 1987.
  5. Ford, 1987, op. cit.
  6. John M. McCann and John P. Gallagher, Expert Systems for Scanner Data Environments, Boston: Kluwer Academic Publishers, 1990.
  7. McCann, 1986, op. cit.
  8. Management Decision Systems, How to Use Your Express Marketing Decision Support System, February 1983.
  9. John A. Quelch, Paul W. Farris, and James Olver, "The Product Management Audit: Design and Survey Findings", Journal of Consumer Marketing, Vol. 4, No. 3, Summer 1987.
  10. John A. Quelch and Alice MacDonald Court, "Procter & Gamble (A)", Case No. 9-584-047, Boston: Harvard Business School, 1984.
  11. George S. Yip and C. P. Brainard, "Planning at Pepsi (B), The Stategic and Marketing Planning Process," Case No. 9-583-051, Boston: Harvard Business School, 1983.
  12. John U. Farley and Donald R. Lehmann, Meta-Analysis in Marketing: Generalization of Response Models, Lexington, MA: Lexington Press, 1986.