CHAPTER 2 Alternative Approaches to Marketing Intelligence We stated in Chapter 1 that the fundamental purpose of marketing research is to help marketing managers and other business people make decisions they face each day in their areas of responsibility. As directors of their firms' marketing activities, marketing managers have an urgent need for information or marketing intelligence—they might need to know about the changes that could be expected in customer purchasing patterns, the types of marketing intermediaries that might evolve, which of several alternative product designs might be the most successful, the shape of a brand's demand curve, or any of a number of other issues that could affect the way they plan, solve problems, or evaluate and control the marketing effort. Marketing research is traditionally responsible for this intelligence function. As the formal link with the environment, marketing research generates, transmits, and interprets feedback regarding the success of the firm's marketing plans and the strategies and tactics employed to implement those plans. Three main approaches are used to provide marketing intelligence: marketing research projects, marketing information systems (MISs), and decision support systems (DSSs). Projects are completed to address a specific, timely marketing question. MISs and DSSs are used for more continuous monitoring of consumer and market behavior. MISs are usually narrower in scope than DSSs and better for addressing narrow, well-defined, predictable issues; for example, scanner data from groceries may be checked by manufacturers to obtain hourly market shares, if so desired. DSSs are broader in scope than MISs, developed with much consideration of consumer behavior. Once developed, they are popular because they allow a manager to answer simple questions to a user-friendly interface. Then the computer takes the input to a model and produces some numbers and answers, such as sales forecasts for a new product launch. In this chapter, we will describe each of these three approaches as clear and distinct from one another, though in practice, one often blends into another. For example, MISs were the forerunners of DSSs, and many contemporary systems have elements of each. Marketing research projects are used to help construct the modeling within a DSS and are used regularly to supplement it. Further, the increased popularity of large consumer databases (for example, from loyalty programs) is blending the continuity of the available data with its periodic use in projects. The emphasis in this book is on marketing research projects. A project is comprised of steps to be taken to solve a specific problem faced by a marketing manager. The next chapter gives an overview of these steps, and the remainder of the book discusses each one in detail. In this chapter, we provide some appreciation for the differences between projects and the alternative schemes of MISs and DSSs and their complementary roles in providing marketing intelligence. First, the philosophical difference is explained. Then we discuss the essential nature of MISs and the components of DSSs. A critical element to any marketing research project, MIS, or DSS is the data set. Because these data sets are increasingly becoming huge, we will discuss issues of "data mining." Finally, the pervasiveness of data collection on consumer and market behavior regularly raises ethical issues such as privacy, which we also address. Philosophical Difference—Periodic Projects and Continual Intelligence The difference in perspective between a project approach to research and an emphasis on information systems was highlighted years ago in a useful analogy that compares a flash bulb and a candle: The difference between marketing research and marketing intelligence is like the difference between a flash bulb and a candle. Let's say you are dancing in the dark. Every 90 seconds you are allowed to set off a flash bulb. You can use those brief intervals of intense light to chart a course, but remember everybody is moving, too. Hopefully, they'll accommodate themselves roughly to your predictions. You might get bumped and you may stumble every so often, but you can dance along. On the other hand, you can light a candle. It doesn't yield as much light but it's a steady light. You are continually aware of the movements of other bodies. You can adjust your own course to the courses of others. The intelligence system is a kind of candle. It's no great flash on the immediate state of things, but it provides continuous light as situations shift and change.1 Thoughtful marketing managers will recognize the value in conducting marketing research and will commission such projects recurrently. Myopic managers will devise a marketing research project only in times of crisis to be carried out with urgency, which often leads to an emphasis on data collection and analysis instead of the development of pertinent, actionable information. One suggestion for making research information more actionable is to think of management in terms of an ongoing process of decision making that requires a flow of regular input. Both MISs and DSSs represent ongoing efforts to provide pertinent decision-making information to marketing managers on a regular basis. Marketing research projects provide the regular input and can discretely fine-tune these more continuous systems. Marketing Information Systems The earliest attempts at providing a steady flow of information focused on the marketing information system (MIS), which was defined as "a set of procedures and methods for the regular, planned collection, analysis, and presentation of information for use in making marketing decisions."2 The key word in the definition is "regular," since the emphasis in MISs is the establishment of systems that produce information needed for decision making on a recurring basis rather than relying on periodic research studies. The design of MISs begins with a detailed analysis of the decision makers who will be using the system in order to secure an accurate, objective assessment of their decision-making responsibilities, capabilities, and styles. MIS analysts determine the types of decisions to be made and the types of information required to make those decisions. They examine the types of information the individual receives regularly and the special studies that are needed periodically. They seek feedback from the decision makers for improvements to the current information system. Given these specifications, systems designers then attempt to specify, get approval for, and subsequently generate a series of reports that would go to the various decision makers. To complete these tasks, systems designers need to specify the data that would be input to the system, how that data could be secured and stored, how the data in separate data banks would be accessed and combined, and what the report formats would look like. Only after these analysis and design steps are completed can the system be constructed, which is essentially a programming task. Programmers write and document the programs, making data retrieval as efficient as possible in terms of use of computer time and memory. When all the procedures are debugged so that the system is operating correctly, it is put online. Once online, managers with authorized access ask for their reports. The information systems department issues a hard copy or, increasingly, managers simply access the reports directly through their company's intranet via the computer terminals on their desks. When they were first proposed, MISs were held up as an information panacea. The reality, however, often fell short of the promise. The primary reasons are as much behavioral as technical. People tend to resist change, and with MISs, the changes are often substantial. Many decision makers are reluctant, for example, to disclose to others what factors they use (for example, using one's own intuition) and how they combine these factors when making a particular decision. (Some research in cognitive psychology suggests that even when decision makers have no motivation to withhold such information, they are frequently simply inaccurate in their implicit understanding of their own decision-making process.) Without accurate disclosure, it is next to impossible to design reports that will give these people the information they need in the form they need it. Even when managers are willing to disclose their decision-making calculus and information needs, there are problems. Different managers emphasize different things and consequently have different data needs; few report formats can be optimal for different users. Either the developers have to design "compromise" reports that are satisfactory but not ideal for any single user, or they have to engage in the laborious task of programming to meet each user's needs. Moreover, the costs and times required to establish such systems are often underestimated, due to underestimating the size of the task, changes in organizational structure or key personnel, and the electronic data processing systems they require. By the time these systems can be developed, the personnel for which they are designed often have different responsibilities, or the economic and competitive environments around which the systems are designed have changed. Thus, they are often obsolete soon after being put online, meaning that the whole process of analysis, design, development, and implementation has to begin anew. Another fundamental problem with MISs is that the systems do not lend themselves to the solution for the kinds of problems managers typically face. Many of the activities performed by managers cannot be programmed, nor can they be performed routinely or delegated to someone else, because they involve personal choices. Because a manager's decision making is often ad hoc and addressed to unexpected events and choices, standardized reporting systems lack the necessary scope and flexibility to be useful. In addition, some decision making and planning are exploratory, so managers, even if they are willing to, may not be able to specify in advance what they want from programmers and model builders. As decision makers and their staffs learn more about a problem, their information needs and methods of analysis evolve. Furthermore, decision making often involves exceptions to rules and qualitative issues that are not easily programmed. Decision Support Systems As the problems with MISs became more apparent, the emphasis in supplying marketing intelligence on a more regular basis changed from the production of preformatted reports to a decision support system (DSS) mode. A DSS has been defined as "a coordinated collection of data, systems, tools, and techniques with supporting software and hardware by which an organization gathers and interprets relevant information from business and the environment and turns it into a basis for marketing decisions."3 As depicted in Figure 2.1, a DSS is comprised of data, model, and dialog systems that can be used interactively by managers. We discuss each system component in turn. Data System of a DSS The data system in a DSS includes the processes used to capture and the methods used to store data coming from marketing, finance, the sales force, and manufacturing, as well as information coming from any number of external or internal sources. The typical data system will have different modules containing customer information, general economic and demographic information, competitor information, and industry information, including market trends. The customer information module typically contains information on who buys and who uses the product, where they buy and use it, when, in what situations and quantities, and how often. It could also include information on how the purchase decision is made, the most important factors in making that decision, the influence of advertising or some sales promotion activity on the decision, the price paid, and so on. Marketing research projects would typically supply some of the information input to the customer information module of the data system. Other input might come from the purchase of syndicated commercial marketing information (as discussed in Chapter 6). In the section that follows on data mining, we shall see how several firms use their customer information databases. The data module that contains general economic and demographic information attempts to capture some of the most relevant facts about what is happening in the external environment, for example, facts about national or international economic activity and trends, interest rates, unemployment, or changes in GNP. The demographic facts would concern changes in population, household composition, or any of the other factors that could potentially affect the future success of the firm. Much of this input would come from government data, primarily from the various censuses (discussed in more detail in Chapter 6). Another data module could contain information on specific competitors. This information would address questions such as: Who are the competitors and what are their market shares? In which market niches do they operate? What is their percentage of sales by product? What are their distribution methods? Where are their production facilities located? How big are they? What are their goals? What are their unique capabilities? The industry information and market trend data module would contain general information on what is happening in the industry, for example, financial information about margins, costs, research and development (R&D) activities, and capital expenditures. It could represent trends in manufacturing or technology, either with respect to raw materials or processes. The industry module could contain information on new technologies that might affect the production process or create new product substitution capabilities. It would also contain information on marketing trends, such as changing distribution or product consumption patterns. A popular application of DSSs that incorporates many of these different data modules is the forecasting of sales for new products. Data input to these systems includes factors such as developmental resources (such as the number of R&D hours behind the prototype), concept testing (such as the percentage of consumers checking the "very likely" box on a scale of likelihood to purchase), prototype testing (such as the comparable percentages of consumers "very satisfied" with product usage), and marketplace promotion expenditures (such as the advertising budget allocated in millions of dollars).4 One important trend in the development of DSSs is the explosion in the past few years in databases that provide information on customers, competitors, industries, or general economic and demographic conditions. Several thousand databases can now be accessed online via computer, compared to fewer than 900 twenty years ago. Several hundred of these apply to the information needs of business. In addition, following the globalization of business, DSSs are increasingly becoming part of international competitive intelligence data systems. The insights that marketing managers can gather from commercially available databases are almost mind-boggling. They certainly dwarf the possibilities of even five years ago. (See the Appendix to Chapter 6 for an extensive list of sources of data.) As the number of databases has expanded, so too has public concern with the issue of privacy, and if and how people's rights to privacy are being violated in the generation and sharing of these databases. We say more about privacy and ethics in the final section of this chapter. Beyond a sensitivity to such public concerns, however, an important criterion as to whether a particular piece of data should be included in the data bank is whether it is useful for marketing decision making. The basic task of a DSS is to capture relevant marketing data in reasonable detail and to organize that data in a truly accessible form. It is crucial that the database management capabilities built into the system can logically organize the data the same way a manager does. Model System of a DSS The model system includes all the routines that allow the user to manipulate the data to conduct the kind of analyses desired. Whenever managers look at data, they have a preconceived idea of how something works and, therefore, what is interesting and worthwhile in the data. These ideas are called models.5 Most managers also want to manipulate data to gain a better understanding of a marketing issue. These manipulations are called procedures. The routines for manipulating the data may run the gamut from summing a set of numbers to conducting a complex statistical analysis to finding an optimization strategy using some kind of nonlinear programming routine. In the real world, "the most frequent operations are basic ones: segregating numbers into relevant groups, aggregating them, taking ratios, ranking them, picking out exceptional cases, plotting and making tables."6 The explosion in recent years in the number and size of the databases available has triggered a commensurate need for ways to analyze them efficiently. Analyzing these large databases has become known as data mining, which we discuss in the next section. For example, the huge quantities of scanner data that brand managers of consumer packaged goods receive every week require a great amount of time in order for even an astute analyst to provide simple summaries showing the major trends. In response, a number of firms have been developing expert systems, computer-based artificial intelligence (AI) systems that attempt to model how experts process information to solve the problem at hand. Figure 2.2 displays the type of output provided by The Partners, an expert system developed by Information Resources, Inc., for the analysis of this kind of scanner-generated data. The Partners can provide highlights of the performance of a brand and competitors' brands within minutes. It can sort through all the data and provide a comparison of current results with past results, by brand and by category as well as by markets, regions, or key accounts. Moreover, as the figure indicates, the system can even produce a memo highlighting the major findings.7 Dialog System of a DSS—The Interface The dialog systems, or language systems, are most important and differentiate DSSs from MISs. This interface permits managers who are not programmers themselves to explore the databases, using the system models to produce reports that satisfy their particular information needs. The reports can be tabular or graphical, and the report formats can be specified by individual managers. The interfaces are often menu-driven to facilitate userfriendliness, reduce errors, and increase usage. Instead of funneling their data requests through a team of programmers, managers can conduct their analyses themselves, which allows them to target the information they want rather than being overwhelmed with irrelevant data. Managers can ask a question and, on the basis of the answer, follow it with a subsequent question, and thus proceed interactively. As the availability of online databases has increased, so too has the need for better dialog systems. The dialog systems are what put data at the managers' fingertips. That sounds simple enough, but it is a difficult task because of the large amounts of data that are available, the speed with which the data hit a company, and the fact that data come from various sources. To compound matters, the geographic boundaries used by the data suppliers differ from each other and most often from the firm's own geographic territories. Further, the services typically collect data on different time cycles. Some might provide it weekly; others might provide it twice a month, monthly, or even less often. The discrepancies must be reconciled in a meaningful way if the various data are going to be combined in a way that enables effective decision making. A relatively new way to handle these problems is through distributed network computing. These systems rely on computers that are linked together. Because the computers are linked, users do not have to be concerned about where the information is stored in the network. More important, the systems used to access and manipulate the data use a common interface or server. Through that server, the analyst can enter data; query data; do spreadsheet analyses, plots, or statistical analyses; or even prepare reports, all through some very simple commands (see Figure 2.3). This structure is likely to continue because firms are finding that decentralization has other benefits, such as greater protection from hackers. The Internet is also affecting the way businesses handle marketing intelligence. This global network was once limited to academicians and government employees sharing technological information, but it now extends to include computer servers at businesses and access providers. The Internet has grown explosively in terms of the number of users and the kinds of information available. The Internet now links people from more than 150 countries, with 100 million users in the U.S. alone. Its popularity is due in part to its easy access; all that is needed is a personal computer with the right software and modem or Ethernet hookup, plus a service provider, such as a commercial online service (America Online, for example). Many Internet users browse the World Wide Web (WWW), a hypertext system that allows users to receive text, graphics, video, and sound. Hypertext is a method for linking and displaying text and graphics that permits users to click on particular words and images, thereby jumping to related documents or images. The Web's hypertext links may send users from the documents of one organization to those of another, perhaps in another part of the world. The number of Web sites has grown exponentially and is no longer estimable. Online penetration is approximately 40 percent in the U.S. (100 million individual users). Average weekly online time is 7.1 hours, just half of the time spent watching television (15.6 hours). Predominant online activities are exchanging e-mail (in which 96 percent of users engage at least once a month), using search engines (88 percent), and researching products and services (72 percent). 8 Many of the new Web sites contain information that belongs in companies' data systems, and the use of the Web to access information on customers, competitors, industries, or general economic and demographic conditions will continue to grow rapidly.9 Moreover, the search engines that have been designed to search the Web have proven to be so useful that many companies have begun using them to manage their intranets, or internal data systems. Examples of DSSs DSSs have long allowed managers to play "what-if" scenarios. Managers can simulate real-world conditions to learn about the relationships among different marketing actions and likely competitive response in a virtual marketplace, without the risk that such experimentation would cause in the real marketplace. Simulations can be programmed to be as complex and realistic as there are variables, data, experience, and assumptions to help support the interrelationships among the elements of the simulation. Input can include data and estimates on market characteristics (such as seasonality), customer preferences (such as attributes sought per segment, price sensitivity), customer loyalty (such as switching costs, trends over time), competitors' positions (such as consumers' judgments of their known product attributes, their prices and perceived prices), and competitors' access to customers (such as their distribution channels). Good simulations allow users to modify the input assumptions and parameters to see the range in possible outcomes, so as to make predictions with some degree of confidence. Shell Oil Products Co. recently used such a simulation and nixed plans to build unmanned service stations (that is, with self-service gas pumps only) because its analysis indicated that competitor reaction would be swift and sure, and any benefits due to the stations would be minor and short-term. Shell estimates that not investing in the self-serve stations saved them some $29 million.10 DSSs are increasingly relying on more sophisticated AI modeling and programming. Such systems still begin with the kind of information that marketing research projects can provide, including customers' historical behavior (for example, purchase transactions, reactions to past promotional efforts), and perceptual data (from surveys and internal databases). Banks use decision tree AI tools to analyze a database of loan applicants to predict who will pay back a loan and who will not. Such systems usually begin by making classifications on current customers for whom the outcome is known (that is, whether they paid back the loan or not). Decision tree analysis detects which customer characteristics are most diagnostic in distinguishing the good bets from the deadbeats. One bank found that home ownership enhances the likelihood of payback over rental status by a wide margin (defaults were 21.4 percent for home owners and 77.3 percent for renters). But there were differences even within these categories. For example, renters who had insurance defaulted less (51 percent) than their uninsured counterparts. Decision trees create rule-based procedures for the classification of the new loan applicant: if the loan applicant is a renter, for example, then default probability is 77.3 percent; if the loan applicant rents and the renter owns insurance, default probability drops to 51 percent. In developing the decision tree rules, the model iterates toward better prediction, and the iteration gives the appearance that the machine is "learning," hence the label "artificial intelligence."11 DSSs can be used for internal customers as well. Home Depot carries an extensive inventory in more than 400 retail outlets. Their stores have radio frequency–transmitted links to their centralized corporate data warehouse, which enables a real-time inventory and guides the store managers in placing orders. The firm estimates that the system saves it the costs of one administrative employee per store and gives the current employees more time to assist customers.12 Comparing DSSs, MISs, and Marketing Research Projects In comparing DSSs and MISs, we note that both are concerned with improving information processing so that better marketing decisions can be made, yet they also differ in a number of ways: 1. DSSs tend to be aimed at the less well-structured, under-specified problems that managers face rather than at those problems that can be investigated using a relatively standard set of procedures and comparisons. 2. DSSs attempt to combine the use of models, analytical techniques, and procedures with the more traditional data access and retrieval functions. 3. DSSs specifically incorporate features that make them easy to use in an interactive mode by nontechnical people, including such things as menu-driven procedures for doing an analysis and graphical display of the results. Regardless of how the interaction is structured, DSSs respond to users' requests in "real time," that is, when the request is made so as to be timely in decision making. 4. DSSs emphasize flexibility and adaptability. They can accommodate different decision makers with diverse styles as well as changing environmental conditions. In comparing DSSs and the traditional marketing research project approach, we note that the explosion in databases and DSSs has only increased the need for traditional marketing research projects and for understanding their strengths and weaknesses in gathering marketing intelligence. DSSs and projects are not competitive mechanisms for marketing intelligence, but are complementary and function best if they are well integrated. 13 For one thing, many of the project-oriented techniques discussed in this book are used to generate the information that goes into the databases that businesses use in their DSSs. Thus, the value of the insights gained from these databases depends directly on the quality of the underlying data, and users must be able to assess that quality. In addition, although a DSS provides valuable input for broad strategic decisions, allows managers to stay in tune with what is happening in their external environments, and serves as an excellent early warning system, it sometimes does not provide enough information about what to do in specific instances. Examples include when the firm is faced with a new product introduction, a change in distribution channels, and the effectiveness of a new promotion campaign. When actionable information is required to address specific marketing problems or opportunities, the research project continues to play a major role. In sum, both DSSs and marketing research projects are approaches to marketing intelligence that can be expected to remain important. In an increasingly competitive world, information is vital, and a company's ability to obtain and analyze information will largely determine the company's future. The light from both flash bulbs and candles is necessary. Data Mining Analyzing large databases has become known as data mining, and businesses hope it will allow them to boost sales and profits by better understanding their customers. The analysis of databases is not new—what is new and challenging is the extraordinary size of these databases. The availability of huge databases began with scanner purchase data (discussed in Chapter 6). Estimates suggest that marketing managers in packaged goods companies are inundated with 100 to 1,000 times more bits of data than even a few years ago because of the adoption of scanner technology in their channels of distribution. Some data mining techniques also arose in response to "database marketing" or "direct marketing" (for example, by catalogue vendors or coupon distribution providers) in which a company is trying to form relationships with its individual customers, as marketing attempts to proceed from "mass" (one media message for all potential buyers) to "segments" (some targeting and positioning differences) to "one-to-one" marketing. In order to achieve such tailored market offerings, a company has to know a lot about its customers—hence the data contain many pieces of information on each of the company's many customers. Traditionally, a company's database would have contained only current business information, but many now contain historical information as well. These "data warehouses" literally dwarf those available even a few years ago. For example, WalMart has contracted with NCR Corporation to build a data warehouse with 24 terabytes (1 terabyte = 1,000 gigabytes) of data storage, which will make it the world's largest data warehouse. The system will provide information about each of Wal-Mart's over 3,000 stores in multiple countries. Wal-Mart plans to use the information to select products that need replenishment, analyze seasonal buying patterns, examine customer buying trends, select markdowns, and react to merchandise volume and movement. 14 See Research Realities 2.1 to gain a better understanding of the size of a number of different companies' customer information databases. In response to the increasingly massive data sets, firms have been working to create increasingly sophisticated data mining technologies (hardware and software) to analyze the data. Data mining uses massively parallel processing (MPP) and symmetric multiprocessing (SMP) supercomputer technologies (during which multiple data points and subroutines may be processing simultaneously, compared with old-fashioned "serial" processing, in which one datum is processed after another). These huge machines support "relational" database programs that can slice massive amounts of data into dozens of smaller, more manageable pools of information. Sometimes these intensive approaches are applied to databases that are being analyzed with fairly traditional statistical techniques. For example, regression (see Chapter 16) is still a premier analytical tool, because many predictors can be used to capture complex consumer decision-making and market behavior—forecasting sales as a function of season, price, promotions, sales force, competitor factors, and delivery delays.15 Other popular techniques of data mining include cluster analysis for segmentation and neural networks (see Chapter 17 for these multivariate statistical methods).16 Businesses regularly use data mining analytical tools to mathematically model customers who respond to their promotional campaigns versus those who do not. The effects of direct mailing efforts, for example, are easily measured and compared as a function of customer information (demographics such as age, household size, income) and purchase behavior (past buying history, cross-sales).17 Data mining can also be used to measure incremental business (additional traffic, sales, profits) that may be directly attributed to a recent promotion by deliberately withholding the promotional mailing from a "control" group (these "experimental" techniques are discussed in Chapter 5). 18 In addition to standard techniques being applied to these huge data sets, marketing research methodologists are creating techniques and software especially for data mining analyses on large data sets. Sales of such customer management software are currently growing at five times the rate of the overall software market, as managers struggle to track every encounter with each customer, to facilitate call-center interactions between customers and customer service representatives, and to manage internal customers, for example, one's sales force. 19 Some of these relational database systems include NCR's Teradata system for Unix or Windows NT machines, IBM's Intelligent Miner, and SAS's Enterprise Miner.20 Other software companies offer "content aggregator" services that synthesize multiple databases—company financial information, histories, executive profiles, and the like. 21 As an illustration of a data mining exercise, Farmers Insurance used IBM's DecisionEdge software to look at the 200 pieces of information the company maintained on its database of 10 million automobile insurance policy owners. Think of a sports car owner and "you probably imagine a twenty-something single guy flaming down the highway in his hot rod." This profile fit many of its customers, but the data mining exercise identified another segment of sports car owner—married baby boomers with kids and more than one car. These customers produced fewer claims, yet had been paying the same sports-car surcharge. With this information in hand, Farmers could charge them less, providing greater value and customer satisfaction.22 For additional examples of the kinds of consumer insights gained from data mining, see Research Realities 2.2. There is no question that the explosion in databases, computer hardware and software for accessing those databases, and the World Wide Web are all changing the way marketing intelligence is obtained. Not only are more companies building DSSs, but those that have them are becoming more sophisticated in using them for general business and competitive intelligence. This, in turn, has produced some changes in the organization of the marketing intelligence function. One change has been the emergence of the position of chief information officer, or CIO. The CIO's major role is to run the company's information and computer systems like a business.23 The CIO serves as the liaison between the firm's top management and its information systems department. He or she has the responsibility for planning, coordinating, and controlling the use of the firm's information resources and is much more concerned with the firm's outlook than with the daily activities of the department. CIOs typically know more about the business in general than do the managers of the information systems department, who are often more technically knowledgeable. In many cases, the managers of the information systems department will report directly to the CIO. Information systems are not intended to be simply data warehouses—the management of information is ideally designed as an electronic library that allows all employees access to the "firm's collective wisdom."24 Privacy and Other Ethical Issues As the number of databases has expanded, so too has public concern with the issue of privacy, and if and how people's rights to privacy are being violated in the generation and sharing of these databases. For example, motor vehicle bureaus sell information, including a driver's name, address, height, and weight, sometimes for relatively innocuous uses, such as targeted marketing lists (Sears, Roebuck & Co. has used state records on height and weight to pick prospects for its "Big and Tall" men's catalog, for example), other times to private investigators seeking to locate everyone from criminal defendants and witnesses, to fathers who are delinquent in child support. The generation and sales of Department of Motor Vehicle (DMV) data sets continues to be controversial; part of the Brady Handgun Act prohibited the release of such personal information. However, since the Supreme Court ruled this prohibition unconstitutional, DMV data will continue to be used and misused, at least for the near future.25 The sharing of financial data, personal information, and consumer purchases has generated much controversy. There is little doubt that as the ability to gather and organize individual-level data expands, so will the controversy regarding individual versus company rights. DoubleClick, the Internet advertising pioneer and the largest of the Web ad servers, has been targeted by privacy advocates for a variety of actions, including their nondisclosure of "cookie" file placements and surreptitious collection of data, their plan to merge lists of individual consumer identities with lists of their Web activities, and their unresponsiveness to critics. 26 For more information on cookies and related issues, see Research Realities 2.3. There is concern about how the consumer purchasing and Web-surfing data will be used, including an electronic version of "redlining," a practice of crossing out certain members of a list (that is, a customer database) as being poor sales prospects (meaning unlikely to be profitable). Redlining was deemed unacceptable because it was based on stereotypes about people living in certain geographical locations. The electronic classification is presumably built on data profiles, but clearly has the potential for misuse, if managers make superstitious assumptions based on the Web sites you visit, the books you buy, your mortgage status, and the like. Current examples of data used to assist decisions include: Visa International managers watch a customer's behavior to spot fraud and identify people who may go bankrupt First Union Bank sorts customers into value segments on which customer service is based (for example, flexibility on credit card rates) Catalina Supermarkets offer free home delivery exclusively to its most profitable customers 27 Companies planning on entering particular types of data in their data systems need to be sensitive to privacy issues.28 Research Realities 2.4 offers a privacy checklist companies might use when developing their databases. Marketing Research—Ethics Beyond Privacy Privacy is a particular concern with data mining and the existence of huge databases, and the issues are being exacerbated by the pervasiveness of the Internet. However, marketing researchers have long recognized that they must take care in conducting their business in a professional manner. In this final section, we discuss broader ethical issues beyond those of privacy and Internet databases. Marketing researchers need to recognize that the effective practice of their profession depends a great deal on the goodwill of and participation by the public. In addition, while the current discussions in the media about privacy issues revolve primarily around the Internet (such as sharing of consumer data, Web site visits, and personal information), the marketing researcher is also affected by the American public's greater protectiveness of its privacy—it is more difficult and costly to approach, recruit, and survey participants. Thus, moral fairness and self-preservation dictate that marketing researchers develop a sense for ethical issues—good ethics is good business. Table 2.1 contains the marketing research code of ethics for the American Marketing Association. In particular, the marketing researcher can encounter ethical issues with three constituencies: (1) the research participants, (2) the client for whom the research is being conducted, and (3) the research team itself. We highlight some of the issues for each relationship in turn. Research Participants Regarding the research participants, there are two main issues: preserving the participants' anonymity and obtaining their consent to participate in the study. Maintaining their anonymity ensures that their identity is safe from invasions of privacy. Information obtained by marketing researchers can be extremely useful to other agents (for example, in compiling mailing lists of likely sales prospects), but unless consumers agree ahead of time to have their identities disclosed, such information should not be passed along. (Internet privacy critics refer to this as the ability to "opt in" or "opt out.") Obtaining a respondent's "informed, expressed consent" is often straightforward; in a mall-intercept, for example, a consumer is confronted in a shopping mall and asked whether he or she would be willing to spend a few moments answering questions. The individual can agree or walk away. However, some common marketing research procedures exist that can involve consumers without their knowledge and therefore without their consent. Participant observation (discussed more in Chapter 7) is the name given to procedures in which the researcher participates in the activity of interest in order to observe people's behavior in their natural environment. A relevant example would be a marketer living among and traveling with a Harley-Davidson gang for six months to study the members' behaviors and consumption patterns. Minimal ethics require the researcher to reveal one's true identity and purpose once the data have been collected and to allow the people who have been observed to read the final report on their activities. Observing people in public places (also discussed in Chapter 7) is less intrusive but common among marketing researchers; it is helpful to watch shoppers' reactions to new floor displays in a store. For many researchers, any activity or conversation occurring in a public place is fair game and arouses no ethical scruples. Strictly speaking, however, the research participants are being involved without their knowledge or consent, and their rights are being infringed. Finally, in some field experiments, benefits are withheld from control groups (see Chapter 5). While this issue is perhaps more critical in medical testing, it can still affect the marketer (such as in studies on sales force management that deprive a control group of incentives). Clients Regarding the client, the marketing researcher must be careful to maintain confidentiality and technical and administrative integrity. Discretion and confidentiality are obligated in not revealing one client's affairs to another client who is a competitor, and in some circumstances, in not revealing the sponsor of the research to participants. Violations of research integrity can range from designing studies without due care through the unnecessary use of complex analytical procedures, to the deliberate fudging of data. It cannot be emphasized enough that in this stage of the profession's development, researchers must maintain the strictest technical integrity if they are to have credibility as professional experts. It is not only unethical but also shortsighted as a business to take advantage of the client's lack of expertise in research design and methodology, because where trust fails, funding eventually does also. Specific recommendations include choosing the simplest appropriate methodology, as opposed to unnecessarily sophisticated and costly techniques; expressing oneself in simple and generally accessible language rather than in intimidating jargon; making explicit mention of the limitations of the research; and refusing any project in which personal problems or conflicts will lead to inadequate performance. Administrative integrity covers issues such as refraining from passing on hidden charges to the client. The Research Team Finally, there are considerations regarding the research team itself. For example, when subordinates are acting according to instructions, the supervisor is partly responsible for their ethical conduct. Moreover, in addition to the official hierarchy, an unofficial sphere of influence exists that renders every team member partially responsible for the others' moral behavior. In particular, studies of organizational culture (such as in marketing research firms) show that actions of top management have been found to be the best predictor of perceived ethical problems for marketing researchers. The source of the boss's influence probably resides in subordinates' fear of reprisals for not conforming and in their acceptance of legitimate authority. As a consequence of the poor examples they see, marketing practitioners do not see themselves as being under pressure to improve their own ethics. See Research Realities 2.5 to get a better understanding of the kinds of ethical issues that confront marketing researchers. Summary This book takes a project-based approach to the provision of marketing intelligence. The difference between a project emphasis to research or the alternative MIS or DSS emphasis is that both of the latter rely on the continual monitoring of the firm's activities, its competitors, and its environment, whereas the former emphasizes the in-depth study of some specific problem or environmental condition. An MIS was defined as a set of procedures and methods for the regular, planned collection, analysis, and presentation of information for use in making marketing decisions. The thrust in designing an MIS is the detailed analysis of each decision maker who might use the system in order to secure an accurate, objective assessment of each manager's decision-making responsibilities, capabilities, and style, and, most important, each manager's information needs. Given the specifications for information needs, system support people develop report formats and efficient systems for extracting and combining information from various data banks. Although MISs did provide more regular marketing intelligence than had been true when firms relied on marketing research projects, such systems suffered from other problems. They required managers to disclose their decision-making processes, which many managers were reluctant to do. Further, the report formats were typically compromises that tried to satisfy the different styles of the different users. And the development time required for these systems often meant that they quickly became obsolete. DSSs are replacing MISs in many companies. A DSS is a coordinated collection of data, systems, tools, and techniques with supporting software and hardware by which an organization gathers and interprets relevant information from business and the environment and turns it into a basis for marketing action. A DSS concentrates on the design of data systems, model systems, and dialog systems. The data systems include the processes used to capture and store information useful for marketing decision making. A marketing research project might be one input to a data system. The model system includes all the routines that allow users to manipulate data to conduct the kinds of analyses they desire. The dialog systems are most important and most clearly differentiate DSSs from MISs. They allow managers to conduct their own analyses while they or one of their assistants sit at a computer terminal. This allows managers to analyze problems using their own personal insight into what might be happening in a given situation, relying on their intuition and experience rather than on a series of prespecified reports. This not only eliminates a lot of irrelevant data, but also saves time because managers can program the analysis themselves rather than waiting for the computer department to process their request for some specific information. Data mining is the term used to describe the analysis of huge consumer databases—scanner purchase data, or investigations in database marketing or direct marketing—tasks for which companies can have hundreds of pieces of information on each of its millions of customers. Large databases require special storage and increasingly sophisticated hardware and software to enable massively parallel processing and symmetric multiprocessing. Sometimes traditional statistical techniques, such as regression or cluster analysis, may be applied to these huge data sets, but increasingly, special customer management software is used, especially for data mining. Information management is a challenge becoming recognized in the appointment of CIOs—chief information officers—whose guidance in the information assets of a firm should only increase in the future, particularly as the World Wide Web allows for greater access to more data. In this chapter, we also considered a number of ethical issues that the marketing researcher must face, beginning with the acknowledgment of privacy issues in reaction to the explosion of the number and size of consumer information databases. Guidelines were offered to respect consumers' concerns for privacy, in which policy disclosure and consumer choice (for example, to "opt out") figure prominently. In addition, for the ideal purpose of professionalism and the pragmatic purpose of sustained business, marketing researchers are encouraged to consider additional ethical issues beyond privacy. These broader ethics concern the treatment of the research participants (such as preserving their anonymity and obtaining their consent), the research client (maintaining confidentiality among competing clients, not commissioning excessive analyses, charging fairly), and the research team itself (sharing responsibility for one another's behaviors and the importance of ethical leadership). Questions 1. What is a marketing information system? How does a project emphasis to marketing research differ from an information systems emphasis? 2. 3. 4. 5. 6. 7. 8. What are the steps in MIS analysis? In developing an MIS system? What are the main differences between an MIS and a DSS? In a DSS, what is a data system? A model system? A dialog system? Which of these is most important? Why? What is (are) the likely future approach(es) to marketing intelligence? Will there be a change in the relative importance of traditional research and MISs and DSSs? What is data mining? How does it differ from traditional data analysis? What are the central privacy issues of consumer data? How does the Internet impact these issues? What classes of ethical issues does the marketing researcher face, beyond privacy? Applications and Problems 1. 2. 3. 4. You are responsible for deciding whether to adopt an MIS or a DSS for the following situations. Which system approach would you choose? Why? a. Production of profit and loss statements to estimate customer lifetime values for segments of video renters. b. Introduction of a new product line extension for Smucker's preserves and jellies. c. Determination of seasonal pricing schedules for Johnson outboard motors. d. Identification of the amount of time spent on hold by consumers on a toll-free, customer-service assistance telephone line. Consider the industry of healthcare management, consulting, or financial investments. What specific capabilities of a DSS would enable greater customer satisfaction and profitability? What kinds of input should your company seek? Who should use the system, and to address what specific needs and questions? You are the vice president of international marketing for a consumer packaged-goods company. In a recent board of directors meeting, it was decided that you would head the development of a competitive information system (CIS) for your organization. You have been asked to write a brief description of the types of data to be stored in the CIS along with possible uses of the data by employees within your company. Write a clear and concise paragraph describing your recommendations. Imagine that you work for an airline. Your company has an extended database of customers' travel records. For the customers who are frequent flyer members, you also have personal data, both data that they have given freely (for example, when applying for membership) as well as data from other sources (such as Web surfing) that you can attach to your members' records because the identification information is clear. You and your data analyst have been mining this large data set to get ideas of travel packages that you could offer that might interest these flyers. You notice that a good portion of your young, male business class flyers who report using their laptops in-flight are single. You consider offering a discount to travel destinations that specialize in singles. You think about sending notice of this promotion via e-mail. You also notice that a disproportionately large number of flights on which more than 5 percent of the seat occupants are children tend to be costly in terms of amenities, post-flight cleaning, and flight attendant time-spent, and these flights tended to generate relatively more customer complaints. You consider increasing fares for children. Do either of these issues raise privacy or ethical concerns? Endnotes 1Statement by Robert J. Williams, who was the creator of the first recognized MIS at the Mead Johnson division of the Edward Dalton Company. "Marketing Intelligence Systems: A DEW Line for Marketing Men," Business Management (January 1966), p. 32. 2Peter D. Bennett, ed., Dictionary of Marketing Terms, 2nd ed. (Chicago: American Marketing Association, 1995), p. 167; also see Berend Wierenga and Gerrit H. van Bruggen, "The Integration of Marketing Problem-Solving Modes and Marketing Management Support Systems," Journal of Marketing 61 (July 1997), pp. 21–37. 3Bennett 1995, p. 77. For a general discussion of the design of business intelligence systems, see George M. Marakas, Decision Support Systems in the 21st Century (Upper Saddle River, NJ: Prentice Hall, 1998) and Won Jun Lee and Kun Chang Lee, "A Meta Decision Support System Approach to Coordinating Production/Marketing," Decision Support Systems 25 (April 1999), pp. 239–250. 4Morris A. Cohen, Jehoshua Eliashberg, and Teck H. Ho, "An Anatomy of a Decision-Support System for Developing and Launching Line Extensions," Journal of Marketing Research 34 (February 1997), pp. 117–129. 5John D. C. Little and Michael N. Cassettari, Decision Support Systems for Marketing Managers (New York: American Management Association, 1984), p. 14. See also Efraim Turban and Jay E. Aronson, Decision Support Systems and Intelligent Systems, 5th ed. (Upper Saddle River, NJ: Prentice Hall, 1998). 6Little and Cassettari, Decision Support Systems, p. 15. 7For general discussions of expert systems, see Luiz Moutinho, Bruce Curry, Fiona Davis, and Paulo Rita, Computer Modeling and Expert Systems in Marketing (New York: Routledge, 1994); Hugh J. Watson, George Houdeshel, and Rex Kelly Rainer, Jr., Building Executive Information Systems and Other Decision Support Applications (New York: Wiley, 1997). 8Andrea Petersen, "Lost in the Maze," The Wall Street Journal (December 6, 1999), p. R6. 9For discussions of how to develop an industry overview, for example, using the Internet, see Marydee Ojala, "Industry Overviews: Turning Industry Question Marks into Answers," Online User (July/August 1996), pp. 14–19. For a general discussion of doing marketing research online, see Reva Basch, "A Strategy for Market Research Online," Online User (May/June 1996), pp. 42–43. 10David 11Barry J. Reibstein and Mark J. Chussil, "Virtual Competition," Marketing Research 9 (Winter 1997), pp. 44–51. de Ville, "Intelligent Tools for Marketing Research," Marketing Research 9 (Spring 1997), pp. 40–43. 12Norbert Turek, "Decision in to Action," Informationweek (October 26, 1998), pp. 85–90. 13Vijay Mahajan and Jerry Wind, "Rx for Marketing Research," Marketing Research 11 (Fall 1999), pp. 7–13. Is in the Details," Discount Store News (November 23, 1998), pp. S5–S7. 15Scott Shrake, "Regression Can Be a Good Thing," Target Marketing 22 (October 1999), p. 56. 16Stewart Deck, "Mining Your Business," Computerworld (May 17, 1999), pp. 94–98. 14"Sell-Through 17Patrick Hanrahan, "Mine Your Own Marketing Data," Target Marketing 23 (February 2000), p. 32. 18Robert Bibb, "Measuring Direct Mail Results," Discount Merchandiser 40 (January 2000), p. 94. 19Steve Hamm and Robert D. Hof, "An Eagle Eye on Customers," Business Week (February 21, 2000), pp. 67–76. 20Allan Holbrook, "Teradata Scales a Data Mountain," InfoWorld (February 14, 2000), pp. 81, 90; Beth Davis, "Data Mining Transformed," Informationweek (September 6, 1999), pp. 86–88. 21Hal Kirkwood, "A Global Business Information Newcomer: SkyMinder.com," Online 24 (March 2000), pp. 62–65. 22Jennifer Lach, "Data Mining Digs In," American Demographics (July 1999), pp. 38–45. J. Honomichl, "Why Marketing Information Should Have Top Executive Status," Journal of Advertising Research 34 (November/December 1994), pp. 61–66. For discussion of some of the problems facing CIOs, see Kaushik Shah, "Manager's Journal: The Perils of Techno Hype," The Wall Street Journal (March 25, 1996), p. 14A. 24Robert Sutton, "Knowledge Management Is Not an Oxymoron," Computerworld (January 3, 2000), p. 28. 23Jack 25Michael W. Miller, "Debate Mounts over Disclosure of Driver Data," The Wall Street Journal (August 25, 1992), p. B1. See also Diane K. Bowers, "Research Interests Protected in the Use of DMV Records," Marketing Research: A Magazine of Management & Applications 7 (Winter 1995), p. 45; John Gibeaut, "Keeping Federalism Alive," ABA Journal (January 1998), pp. 38–39. 26"Crisis Rx for DoubleClick," Advertising Age (February 28, 2000), p. 58; Ira Teinowitz and Jennifer Gilbert, "Marketers Address Web Concerns," Advertising Age (February 21, 2000), pp. 1, 60. 27Marcia Stepanek, "Weblining," Business Week E.Biz (April 3, 2000), pp. EB26–34. 28For general discussions of the privacy–database controversy, see Bruce Horowitz, "Marketers Tap Data We Once Called Our Own," USA Today (December 19, 1995), pp. 1A–2A; "How to Safeguard Your Privacy," USA Today (December 19, 1995), p. 48. FIGURE 2.1 Components of a Decision Support System (DSS) FIGURE 2.2 Example Report Produced by the Expert System The Partners FIGURE 2.3 Use of Dialog Systems with Common Server or Interface Using Simplified, Standardized Instructions to Perform Multiple Tasks TABLE 2.1 AMA Marketing Research Code of Ethics The American Marketing Association, in furtherance of its central objective of the advancement of science in marketing and in recognition of its obligation to the public, has established these principles of ethical practice of marketing research for the guidance of its members. In an increasingly complex society, marketing management is more and more dependent upon marketing information intelligently and systematically obtained. The consumer is the source of much of this information. Seeking the cooperation of the consumer in the development of information, marketing management must acknowledge its obligation to protect the public from misrepresentation and exploitation under the guise of research. Similarly, the research practitioner has an obligation to the discipline and to those who provide support for it—an obligation to adhere to basic and commonly accepted standards of scientific investigation as they apply to the domain of marketing research. For Research Users, Practitioners, and Interviewers 1. No individual or organization will undertake any activity which is directly or indirectly represented to be marketing research, but which has as its real purpose the attempted sales of merchandise or services to some or all of the respondents interviewed in the course of the research. 2. If respondents have been led to believe, directly or indirectly, that they are participating in a marketing research survey and that their anonymity will be protected, their names shall not be made known to anyone outside the research organization or research department, or used for other than research purposes. For Research Practitioners 1. There will be no intentional or deliberate misrepresentation of research methods or results. An adequate description of methods employed will be made available upon request to the sponsor of the research. Evidence that fieldwork has been completed according to specifications will, upon request, be made available to buyers of the research. 2. The identity of the survey sponsor and/or the ultimate client for whom a survey is being done will be held in confidence at all times, unless this identity is to be revealed as part of the research design. Research information shall be held in confidence by the research organization or department and not used for personal gain or made available to any outside party unless the client specifically authorizes such release. 3. A research organization shall not undertake marketing studies for competitive clients when such studies would jeopardize the confidential nature of client-agency relationships. For Users of Marketing Research 1. A user of research shall not knowingly disseminate conclusions from a given research project or service that are inconsistent with or not warranted by the data. 2. To the extent that there is involved in a research project a unique design involving techniques, approaches, or concepts not commonly available to research practitioners, the prospective user of research shall not solicit such a design from one practitioner and deliver it to another for execution without the approval of the design originator. For Field Interviewers 1. Research assignments and materials received, as well as information obtained from respondents, shall be held in confidence by the interviewer and revealed to no one except the research organization conducting the marketing study. 2. No information gained through a marketing research activity shall be used, directly or indirectly, for the personal gain or advantage of the interviewer. 3. Interviews shall be conducted in strict accordance with specifications and instructions received. 4. An interviewer shall not carry out two or more interviewing assignments simultaneously, unless authorized by all contractors or employers concerned. Members of the American Marketing Association will be expected to conduct themselves in accordance with the provisions of this code in all of their marketing research activities. Reprinted with permission from the American Marketing Association. RESEARCH REALITIES 2.1 Size of Customer Information Databases Fingerhut is the $2 billion mailer of 400 million catalogs a year to 65 million customers, approximately 10 million of whom are considered "active" customers. Fingerhut stores 1.5 terabytes of data representing transactions, demographics, and psychographics on these customers. A data mining expedition identified increased purchasing by households that had recently moved, and in response, Fingerhut created a special "mover's" catalog to address these special consumers' needs. They estimate their tailoring of their direct marketing efforts saves the company more than $3 million a year. American Century Investments, a provider of mutual funds, stores 800 pieces of information for each of 25 million customers. Segmentation of their customers allows a more refined direct mailing effort, which tripled their customer response to a recent promotional effort. Vermont Country Store sends out 3 million catalogs annually, for yearly sales around $50 million. More effective targeting based on a segmentation of their 10 years of accumulated marketing data enhanced recent sales of different target products from 2 to 12 percent. Hallmark Cards assists its 15,000 store managers in "SKU (stock-keeping unit) optimization," the allocation of store square footage to its 40,000 products. It can determine which cards and gifts are selling on any given day at any given retail outlet. First Union Corp. created a customer data repository that holds two years' worth of transactional information for each of its 16 million customers. The 27-terabyte relational database is mined to deliver optimally appropriate financial products to its customers. Pillsbury's internal network allows its employees in over 70 countries access to data of several kinds, including consumer feedback that has been logged into a massive database (based on 3,500 calls a day to the 800 number printed on every Pillsbury product), manufacturing (testing equipment at new plants, statistics on production quality and packaging), and so on. Any employee, at a plant or at a sales call pitching new products to a grocer, can access the company's data. Sources: Stewart Deck, "Mining Your Business," Computerworld (May 17, 1999), pp. 94–98; Jennifer Lach, "Data Mining Digs In," American Demographics 21 (July 1999), pp. 38–45; "Sell-Through is in the Details," Discount Store News (November 23, 1998), pp. S5–S7; Steward Deck, "Warehouse Expansion," Computerworld (April 19, 1999), p. 16; Roger O. Crockett, "A Digital Doughboy," Business Week E.Biz (April 3, 2000), pp. EB78–86. RESEARCH REALITIES 2.2 Consumer Insights Gained from Data Mining Loyalty cards, such as those offered at supermarket retailers, offer consumers discounted prices and coupon incentives. In the past 10 years, more than 100 million loyalty cards and key tags have been issued: 30 percent of supermarket customers have them; of those, 70 percent use them. Companies know that all customers are not equal, and loyalty cards enable one-to-one marketing, customizing the shopping experience for households with different purchasing profiles (for example, sensitivities to price, value, brand, and quality). Loyalty cards and grocery purchases have yielded consumer insights and marketing actions such as these: 1. Of Diet Coke drinkers, 13% consume 83% of its volume. Taster's Choice is even more extreme—it generates 73% of its sales from only 4% of its customers. 2. Gillette used a direct marketing mailing campaign to send its razors and coupons to men and women who purchased competitors' razors. 3. Veryfine considered changing its fruit drink flavors but met with resistance when its most loyal users indicated that they did not even want the packaging to change. 4. Coca-Cola strengthened its relationship and power with retailers when it demonstrated that customers who purchased Coke as one of the items in their shopping carts were more profitable to the retailer (for the entire basket of purchases) than consumers who did not purchase Coke. Federal Express data mines to obtain customer segments to pinpoint their desires for greater profitability. Customer service representatives are empowered to go to different lengths to satisfy customers who have been segmented as more and less profitable. This customer-relationship management effectively creates a profit-and-loss statement per customer and customer segment. Rubbermaid data mines its warehouse to determine promotional effectiveness. It can model the likely sales resulting from a 25 percent reduction on prices with two-page ads compared to 40 percent price cuts with smaller ads. They also use their data for merchandise optimization and claim that this careful category management also enhances their relationships with their retailers, such as Wal-Mart, Pamida, and Ames. Hotels regularly collect a great deal of information on their guests. They supplement guest history data with guest preferences, and can thereby provide better quality and customized service. Implementers of such data systems find greater customer satisfaction and loyalty, and increased revenue per customer. Sources: Ann M. Raider, "Programs Make Results Out of Research," Marketing News 33 (June 21, 1999), p. 14; Paul C. Judge, "What've You Done for Us Lately?" Business Week (September 14, 1998), pp. 140–146; Clay Dickinson and Maite Tabernilla, "A New Customer Relationship Management Approach," Lodging Hospitality (May 15, 1999), pp. R11–R12. RESEARCH REALITIES 2.3 Web Threats to Privacy—The Cookie Monster There is no question that the Web has the ability to be extremely nosy. Web marketers can determine, without permission, your "domain," the portion of your e-mail address that follows the @ symbol. That can tell marketers you reached their site via a consumer service such as America Online or a corporate connection. Internet marketers can thus target specific domains for their ads. More controversial are the whimsically dubbed "cookies," a technology that allows Web sites to track individual users. Prior to cookies, Web sites could tally requests for information made each day, but they could not tell whether one visitor made 100 requests or 100 unique visitors made one request each. With cookies, a Web site places a file on a visitor's computer that serves as a kind of tracking beacon. The site does not know your name or e-mail address, but it has your IP address and represents you as a distinct user. (Curious? Search your hard drive for a file called "cookies" and open it in a word-processing program to see who has served you a cookie. If you're concerned about privacy, delete the cookie files occasionally.) Cookies feed some of the more dire privacy scenarios. With them, a Web magazine will see which articles you read; a merchant can tell not only which products you bought, but also which product descriptions you simply viewed. (Imagine a supermarket scanner that monitors everything you look at in a store.) Similarly, it not only knows which ads work; it also knows which don't. Still, some of these fears are no doubt overblown. To begin, surfing is an anonymous activity, and cookies cannot automatically penetrate that shield. When you visit a Web site, the computer maintaining that site may know your domain, but it cannot know your identity, or even your e-mail address, unless you volunteer it. Of course, plenty of sites that sell products online require names, addresses, e-mail addresses, and credit-card information. Providing it, however, is your choice. Access to cookie files may become limited in the near future, but not in response to consumers' concerns for privacy. Rather, companies are learning to protect their customers' information from competitors—if Mattel can see where you've surfed, so can Hasbro. Sources: Thomas E. Weber, "Browsers Beware: The Web Is Watching," The Wall Street Journal (June 27, 1996), pp. B10, B12. See also Walter S. Mossberg, "Threats to Privacy On-Line Become More Worrisome," The Wall Street Journal (October 24, 1996), p. B1; Gautam Naik, "Do I Have Privacy On-Line?" The Wall Street Journal (December 9, 1996), p. R12; Kathryn Kranhold and Michael Moss "Keep Away From My Cookies, More Marketers Say," The Wall Street Journal (March 20, 2000), pp. B1, B6. RESEARCH REALITIES 2.4 A Privacy Checklist In the long run, ensuring a customer's privacy can improve a company's profitability. Privacy is really about earning the customer's trust, and trust is a central component of relationship marketing. Anyone who uses personal information to target customers can stay on top of the privacy issue by writing a formal privacy policy. If you need to create a policy or review your old one, here are some basic guidelines: 1. Remember "knowledge, notice, and no." Tell your customers how you will use their personal information. If you plan to share the information with a third party, tell your customers and give them a chance to drop out of the database. Even if you don't sell your customer lists, tell them so. The practice of renting customer lists has become so widespread that a customer may assume you share your lists unless you explicitly tell them otherwise. The adaption of this policy to the sharing and selling of Internet data has been suggested as follows. a. Display your practices—companies must inform consumers how they gather and use information; b. Give people a choice—Web site users can opt in and choose to provide personal information, or they may opt out and choose not to do so; c. Show consumers the data—allow users to view and correct sensitive information, such as financial and medical data; d. Play fair or pay—it has been suggested that some agency such as the Federal Trade Commission (FTC) needs to enforce fair information practices. 2. Exercise conscience and common sense. So far, few legal restrictions apply to the gathering and use of personal information by the private sector. That's why privacy is more about "should" than "must." Clearly, some medical, financial, and lifestyle data are more sensitive than others. Apply a "sniff test" to any proposed reuse of your customer database. Would you be comfortable sending a member of your family the same offers that you propose to mail to your customers? If your company is identified as a sponsor of this mailing, would your 800-number be clogged with complaints? Sources: Mary J. Culnan, "The Privacy Checklist," in Judith Waldrop, "The Business of Privacy," American Demographics 16 (October 1994), p. 55; Heather Green, et al., "Online Privacy: It's Time for Rules in Wonderland," Business Week (March 20, 2000), pp. 83–96. RESEARCH REALITIES 2.5 Marketing Researchers' Own Perceptions of the Difficult Ethical Problems They Face Ethical Issues to Which Practitioners Are Most Sensitive: 1. Maintaining their research integrity: for example, deliberately withholding information, falsifying figures, altering research results, misusing statistics, ignoring pertinent data. 2. Treating outside clients fairly: for example, passing hidden charges to clients, overlooking violations of the project requirements when subcontracting parts of the project. 3. Maintaining research confidentiality: for example, sharing information among subsidiaries in the same corporation, using background data developed in a previous project to reduce the cost of a current project. Factors That Predict Sensitivity to Ethical Issues: 1. Organizational socialization (for example, "I know the rules associated with my job" and "I know what's considered appropriate behavior in my company"). 2. Ability to empathize (such as, "Generally, I find it easy to see things from the other person's perspective"). From John R. Sparks and Shelby D. Hunt, "Marketing Researcher Ethical Sensitivity: Conceptualization, Measurement, and Exploratory Investigation," Journal of Marketing 62 (April 1998), pp. 92–109.