Determining Executive Information Requirements Determining ecutives have little reaction time to make decisions on pricing, product introduction, resource alloca- Executive Information tion, media inquiries, response to competition, and mergers. They need access to information quickly. Requirements: Better, They can not wait several weeks or months for completion of a computer project. Executives and Faster, and Cheaper systems designers need to work together to antici- pate and determine information requirements more accurately. This article addresses four reasons for lengthy information requirements and provides a discussion of the failure to meet these requirements. by Four straightforward solutions are offered to execu- tives to solve this problem.1 James C. Wetherbe The University of Memphis and The University of Minnesota Introduction The new vice president of engineering for an aero- space company noticed that he automatically re- ceived 32 computer-generated reports each month. Executive Summary He was having difficulty determining the purpose of most of them. Convinced his predecessor re- Determining management’s information needs is his- ceived these reports for a good reason and reluctant torically one of the most time consuming and error- to admit his inability to find meaning for them, he prone activities in designing and building informa- had an idea. If he found out who the other manag- tion systems. Most managers spend half their time ers were who also received copies of these reports, obtaining needed information, either informally he could subtly inquire about their use and thereby through meetings, phone conversations, or reading; determine what he was “missing.” or formally through organizational computer-based He asked his assistant to get distribution lists for all information systems. During this process they have 32 reports. Within a week, the assistant returned to sift through a great deal of useless information, with the distribution lists. resulting in “information overload.” The proliferat- ing capabilities and plummeting cost of computers “So, who gets copies?” the vice president should signal relief for weary executives. Unfortu- inquired. nately, most information systems take too long to “Two people—you and me,” the assistant design and build, and then fail to meet executive replied. needs. Indeed, most new systems require extensive Startled, the vice president asked, “What revision (after they are supposedly completed) to do you do with your copies?” even partially fulfill needs, resulting in a terrible “They are just backup copies in case you loss. Most systems are very expensive to develop lose one of yours.” replied the assistant. and even more expensive to revise. As the pace of business accelerates, decisions that used to be made By paying close attention during the next year, the over a period of weeks must now be made in days, vice president determined that he only needed four hours, or even minutes. Failure to provide execu- tives with the needed information in a timely man- ner can result in lost opportunities or in a problem 1 This article draws from earlier work by the not being solved expeditiously. Increasingly, ex- author (Wetherbe, 1991). 1 Determining Executive Information Requirements of the 32 reports. He also determined there was requirements and the remedial actions required, an some very important information he was not get- executive can work with systems analysts to get the ting. He discontinued the unneeded 28 reports, correct system quickly. which resulted in immense appreciation from those who had to provide data to generate them. He then This article covers the causes of and solutions for began to focus on getting the information he really this problem. First, some fundamentals of informa- needed. Over the next year he became frustrated tion and decision making are discussed. Next, the with the inability of the information systems depart- four common mistakes made both by executives and ment to meet his requirements. The information system designers during design of information sys- systems department, in turn, became frustrated with tems are reviewed. Finally, four techniques—cross- his inability to make up his mind about the informa- functional systems, joint application design, struc- tion he needed. Because his requirements were in a tured interviewing, and prototyping—are presented constant state of flux, the department had to con- as pragmatic, easy-to-implement solutions for cor- tinually revise the systems, wasting time and money. rectly determining executive information require- ments. The preceding example is all too common and rep- resents a major source of lost productivity among executives and the organization’s computer person- Managers and Information nel. One of the most important revelations about man- The most important task of an executive is decision agers and information from research and practice is making. Outside of his or her intellect, the most that managers do not know what information they important resource an executive uses is information. need2. A study coordinated by the MIS Research Yet time and again, executives complain they are Center at the University of Minnesota determined overloaded with irrelevant, useless information, that 76 information systems developed in 26 orga- while they are unable to obtain the information they nizations all required minor to major revisions after need in a timely manner. they were completed to even approximate management’s information requirements (Jenkins, The cost and time required to remedy a system that et al., 1984). The systems development process fails to meet management’s needs can often far ex- can be broadly categorized into: (1) designing the ceed the cost and time required to develop the ini- right system and, (2) correctly implementing it. tial system. High as it might be, the revision cost Given an appropriate design, most information sys- might be only a fraction of the opportunity cost of tems departments can successfully implement a sys- management making a bad decision—lack of timely tem. The big problem is correctly determining in- information for decision making might be a more formation requirements and designing the right sys- significant issue. Consider the claims manager who tem. cancelled the insurance of a teenager with a bad driving record. The teenager was the son of the How do most systems analysts go about determin- president of a large corporate account, a fact un- ing what information managers want from their com- known by the claims manager. You can guess the puter system? They do the obvious and the logical. rest of the story. They ask, “What information do you want from the new order processing (or whatever) system?” Fortunately, executives can play a positive role in ensuring that a new system meets their information 2 The inability of managers to recognize infor- requirements and in reducing the time it takes for it mation needs was described by Russell Ackoff to do so. By understanding why systems fail to meet in 1967. 2 Determining Executive Information Requirements Unfortunately, managers usually do not know what project more than a year and double the cost of the information they need. They give it their best at- system. Management was so furious that it cancelled tempt, assuming these brilliant computer wizards the project, which represented a $40 million invest- will sort things out. Several months and millions of ment up to that point, and many information sys- dollars later when the system is delivered, manag- tems people lost their jobs. ers quickly and frequently discover the system does not give them the needed information. Having been victims of management’s inability to properly define information requirements, most sys- Managers ask for changes, and the system analyst tems analysts go to Plan B—the “user sign-off.” This goes into shock. Costs and time to change the de- approach involves asking managers what informa- sign of a system after it is complete are 50 to 100 tion they want from a system and then requiring them times higher than making those same changes dur- to sign a document aimed at contractually obligat- ing the systems design ing them to accept the re- stage. This sounds like an Yet time and again, sulting system. User sign- exaggeration, but it is not. offs have marginal politi- executives complain they If you, or someone you cal value when systems know, has had a house cus- are overloaded with analysts are battling with tom-built, you may be fa- management over system irrelevant, useless miliar with these dynamics. revisions. They, however, For example, consider the information, while they do not solve the functional cost of adding a bathroom problem, which is that after the house is complete are unable to obtain the managers do not know versus the cost of adding a what information they information they need in bathroom during the blue- need. print or design stage. The a timely manner. cost factor explains why so For example, in the bank- many needed revisions are ing illustration discussed never implemented. Consequently, the resulting above, the managers signed off on the design. How- systems are a disappointment. A disappointing sys- ever, once they realized the system would not sat- tem can range from a system that partially fulfills isfy their needs, they blamed the systems analysts management’s requirements (with or without expen- for misleading them. Top management claimed: sive revisions) to one that is totally abandoned, re- “You technical people should have protected us from sulting in a multimillion dollar write-off. our lack of expertise in this area.” A user sign-off is a powerless piece of paper when matched against For example, a major bank recently completed a the fury of top management. multimillion dollar project that tracked all of its cus- tomers’ financial relationships with the institution. Plan C, commonly used by systems analysts, is to The system was demonstrated to various manage- use the “catalog” approach to information require- ment teams who asked if the system would be able ments determination. This involves showing a man- to provide information it was not designed to pro- ager a wide variety of reports, perhaps requested by vide. In other words, managers wanted informa- other managers or available from a commercial soft- tion that had not been requested when the system ware package. The manager reviews these reports, was designed the previous year. Though it would selecting the ones believed to be needed. This may not have been difficult to add the requested capa- seem like a good idea, but it does not work. Offer- bilities had they been identified before systems de- ing managers many reports usually results in requests velopment, adding them now would delay the for them whether needed or not. 3 Determining Executive Information Requirements Research has found that decision makers who are responsible for design, the system carries a very offered cosmetically impressive, but useless reports, strong control orientation as opposed to a general have a high propensity to take them (Benbasat, et management reporting orientation, resulting in bud- al., 1977; Judd, et al., 1981). For example, in one geting systems that end up much like a bank state- study production managers were offered 20 differ- ment. Most people use a bank statement simply to ent reports, eight of which were deemed useful by a reconcile their personal finances, whether on paper panel of production experts. The remaining 12 were or a personal computer. Similarly, many managers useless and were provided solely to see if anyone keep their departmental budgets on a departmental would take them. Uncertain if the information was computer and simply reconcile them with the con- useful, they played it safe and requested it. In prac- trol statement they receive from the budget depart- tice, although the manager requesting the informa- ment. Due to this phenomenon, up to 60 percent of tion does not use it, the manager who replaces him the data entered into personal computers are keyed or her years later may assume the information has from reports generated from other computers in the some value. Consequently, he or she may often be same organization. found staying late at work, sifting through the use- less reports, wasting time, and asking the compel- For example, a marketing vice president for a manu- ling question: “What am I missing from this infor- facturing company wanted to categorize costs and mation used by my predecessor?” Years of this phe- revenue by salesperson, customer, and product. The nomenon result in information overload episodes budgeting system only allocated costs by project similar to the one described in the beginning of this account number. A project could involve more than article. one salesperson, more than one customer, and more Four fundamental mistakes historically have been than one product. Only through extensive data col- made in the process of determining executive infor- lecting from the sales force and the use of spread- mation needs: (1) viewing systems as functional sheet software could the needed information be ob- instead of cross-functional, (2) interviewing man- tained. agers individually instead of jointly, (3) asking the wrong questions during the interview, and (4) not Some would argue that the budgeting department allowing trial-and-error in the detail design process. should incorporate the reporting needs of functional managers into the development of a new budgeting system, but the common argument against this would be increased costs. The flaw in this logic is that the Cross-Functional Systems increase in costs exists anyway because these func- tional managers have to develop their own system, The first mistake in determining information require- resulting in higher costs than if an integrated sys- ments for information systems is viewing the sys- tem was developed and used across functions. tems as functional as opposed to cross-functional (Wetherbe and Vitalari, 1995), a perspective that is As this marketing vice president said, “The account- too narrow and time consuming. Indeed, cross- ing department adds so much overhead to market- functioning is a key principle of cycle time reduc- ing by wanting more and more data categorized in tion (Wetherbe, 1995). For example, when devel- ways that allow them to control and audit us. I need oping a new budgeting system, we tend to focus on information categorized in ways to help us market the information needed by the budget managers or our products effectively and efficiently. The sys- budgeting staff members, rather than all personnel who will use budget information. tems designed by budgeting are not responsive to my needs. Have you ever heard of a manufacturing Virtually all managers need access to budget infor- company that was successful because it had the best mation. Unfortunately, if the budget department is accounting practices in the world?” 4 Determining Executive Information Requirements To illustrate the need to develop systems cross-func- tem, information should be provided to improve all tionally, consider a business process such as cus- decisions. tomer order processing. To process orders, sales- people have to decide on which customers to call For example, consider the last decision listed for and what is available to sell them. Credit depart- the distribution center in Table 1—which custom- ments must decide which customers can have credit ers should receive available inventory. If the dis- and how much, which customers need past-due no- tribution center has five orders but only enough in- tices, and which customers’ credit should be dis- ventory to fill three, a resource allocation decision continued. Distribution centers must decide inven- must be made. Typically, this decision would be tory stocking levels and reorder points, when to made on a first-in/first-out (FIFO) basis, which unload slow moving inventory, and to which cus- would seem equitable and fair given the available tomers limited inventory should be allocated. Ship- information. ping must decide such things as what merchandise to send to which customers, how orders can be Consider the following scenario. Orders for lim- shipped together to save delivery costs, vehicle rout- ited stock arrive in this manner. Customer A orders ing, and vehicle scheduling. These decisions are infrequently, does not need the stock urgently, does summarized in Table 1. In developing a new sys- not pay bills promptly, yields a low profit margin, Table 1. Decision Centers Involved in Order Processing Decision Center Activity Examples of Major Decisions Salespeople Selling • Which major customers to call Merchandise • What to sell customers • What is available to sell Credit Accounts • Which customers should receive credit Department Receivable • How much credit to allow Managment • Which customers need past-due notices • Which customers' credit should be discontinued Distribution Inventory • What inventory to stock Center Management • How much inventory to stock • When to reorder stock • When to unload slow-moving inventory • Which customers to allocate available inventory Shipping Packing and • What merchandise is included in the Department Shipping customer's order Orders • What orders can be shipped together to save delivery costs • Vehicle routing to minimize transportation costs • When vehicle should depart 5 Determining Executive Information Requirements and is on a three-week trucking route cycle. Cus- system and to be sensitive to their decision making tomer B, a high volume buyer, recently complained requirements. Then a system can be developed that of late delivery. He needs the shipment urgently, allows information to flow cross-functionally to im- pays bills promptly, yields a high profit margin, and prove decision making. is located on a frequently used trucking route. In this instance, a shipment made to customer A on a As straightforward as the concept of cross-functional FIFO basis could have disastrous results. systems is, most system analysts attempting to de- velop these systems complain that employees are In trying to improve the quality of the decision, fac- very proprietary about their functional information tors that should be considered include: and are often unwilling to participate in a system that will share information. Recognizing that infor- • How important is each customer to the mation is power, employees are not interested in business? sharing power. • How soon does each customer need delivery For example, sales representatives for a manufac- of the order? turing company were not allowed to access customer credit status. Consequently, they would occasion- • What is the profitability of each order? ally invest substantial effort into obtaining a large order from a client only to have it rejected because • What is the credit status of each customer? of credit/financing conditions. Customer relations were damaged and morale suffered as the sales force • What is the shipping schedule for delivery began to refer to credit as the “sales prevention de- to each customer? partment.” This attitude is totally dysfunctional. Since information is power, the idea is to empower • Has the customer recently been upset because decision makers by giving them the best informa- previous order was late? tion to make these critical decisions. An organiza- tion that does not share information cross-function- Note that the information needed to improve the ally ends up with a situation of the left hand not decision making in the distribution center is gener- knowing what the right hand is doing. ated outside the distribution center. For example, customer need, importance, and profitability are To solve the problem, top management needs to use determined by sales, credit worthiness by the credit its leadership and influence to achieve cross-func- department, and shipping schedule by the shipping tional design. New systems which transcend func- department. tions must have management participation in their design. A very important concept of information manage- ment, therefore, is that most of the information needed to improve functional decision making is generated outside of the function. Sharing infor- Joint Application Design mation within an organization is important in im- proving productivity. When an organization learns The second mistake commonly made in the deter- to share information cross-functionally, employees mination of information requirements is that the sys- are empowered to make better and more productive tems design team usually interviews managers indi- decisions for the organization. vidually instead of using a group process known as joint application design (JAD). As a cycle time prin- The bottom line is that in order to develop a new ciple, this process is referred to as co-locating information system, it is necessary to be aware of (Wetherbe, 1995). The individual interviewing pro- all functions that are impacted by the information cess places cognitive stress on a manager that hin- 6 Determining Executive Information Requirements ders his or her ability to respond adequately to ques- and reducing inventory costs, while those respon- tions. sible for shipping are interested in ensuring good routing of trucks to minimize delivery costs. View- Consider this scenario: A group of strangers come ing the purpose of order processing from a group into your office and ask you to tell 10 jokes. Even perspective would likely result in a design criteria though you probably know 10 good jokes, you might that would focus on achieving prompt, correct de- have difficulty recalling them; most people would. livery of orders to customers while ensuring credit Now, let us change this scenario: A group of you integrity and facilitating good inventory manage- and your fellow managers are put together in a room ment, as well as good routing and scheduling of ship- and asked to generate some good jokes. It is likely ments. you and your colleagues could generate 80, 90 maybe 100 jokes. Each manager would be familiar with perhaps 80-90 percent of them. In other words, It is difficult to achieve this overall perspective if each manager really knew many jokes, but had dif- each manager is interviewed individually. A case ficulty recalling them on an individual rather than study illustrates the need for joint application de- group basis. The moral is that sign from a cross-functional group or collective experi- perspective. A direct mail . . . most of the ences and memory are essen- catalogue company was revis- tial in recalling information. information needed ing its information systems. When people are asked to tell Prior to the cross-functional to improve functional a joke, they generally tell ones design, the credit department they have heard recently. On decision making is viewed its primary goal as en- their own, they can not re- suring payment from custom- member many from the past. generated outside of ers. Ideal performance would Similarly, when managers are be for all customers to make the function. asked what information they all payments—that is, no need, they generally mention credit losses. In an effort to only recent needs, not every- increase its performance in this area, the credit de- thing they need. Therefore, requirements determi- partment had continually requested more and more nation should be done as a group or joint process so information about customers, i.e., credit references, the memory of each manager can be pooled for a credit bureau checks, etc., to the point that credit more thorough job of recalling key information re- costs were becoming excessive. quirements. A second reason for a joint application design is Two mistakes can be committed in making a credit that different functional areas of an organization decision: (1) giving credit to people who will not have different agendas for developing a new infor- pay their bills and (2) not giving credit to people mation system. For example, in the order process- who would pay their bills. After looking at the prob- ing system shown in Table 1, each decision center lem cross-functionally with joint application design, would likely emphasize different design criteria. the organization determined that it was better off Sales may view the primary importance of order pro- not doing any credit checks at all. This rather cessing as ensuring prompt and correct delivery of counter-intuitive conclusion was based upon two key orders to customers. Credit, on the other hand, may understandings that were generated from the joint view the agenda as primarily ensuring that the com- application design. First, the company could send pany gets full payment for all orders. Those respon- catalogues of only low-priced items to first-time cus- sible for inventory management are interested, of tomers. Customers who paid for what they ordered course, in facilitating good inventory management could be upgraded to receive more expensive cata- 7 Determining Executive Information Requirements logues. Those customers who did not pay would be question, it is not at all helpful to managers attempt- dropped from any future mailings. ing to determine what information they need. Sys- tems analysts assume managers surely know what In this way, the company was not inferring whether information they need. The executive assumes the customers would pay from credit reference mate- systems analyst knows what he or she is doing. The rial; based on their own experience they knew which problem is that this technique is akin to a psycho- customers would and would not pay. Their losses analyst talking to a patient and asking, “What type from not receiving payment were less than the costs of therapy do you need?” Or a salesperson acting as for all the credit checks. In other words, the cost of an order taker rather than a problem solver, who unpaid merchandise was the cost of determining asks, “What features do you want?” If patients or whether someone would pay. This information was customers do not know how to look out for them- not only more accurate but also less costly than do- selves, they are unlikely to get satisfactory solutions. ing the traditional credit reference checking. If credit information requirements had been determined with- A personal story illustrates this point. When I moved out considering the context of sales management, to Minnesota, I purchased a home in the country this insight would not have been achieved. with sufficient land to require a tractor mower. I set out to purchase a tractor mower, not knowing much Furthermore, people who would be categorized gen- about tractor mowers, other than that I wanted a Toro erally as higher credit risks have a greater propen- (The dean of our business school was the former sity to purchase from this company’s catalogues. chief executive officer of Toro, and I wanted a Toro Conversely, people who would be considered ex- in my garage just in case he stopped by for a visit). cellent credit risks tend not to buy anything from Consider me a manager who needed to solve a prob- their catalogues. This means the company could lem without knowing specific requirements. send out many catalogues to people with excellent credit ratings but seldom make a sale. Therefore, it When I went into a dealership to purchase a tractor would be losing money by shipping catalogues that mower, the salesperson would ask me what I was never generate any revenue. The company would looking for. I would say I was looking for a Toro be better off marketing to those people who would tractor mower. After that I would find myself in be higher credit risks but not letting them buy any- trouble. The next question I would typically be thing expensive until it was established that the cus- asked was what blade width I wanted. This was not tomer would, in fact, pay for their orders. a question I was really prepared to answer. When I told the salesperson this, I could see him roll his Without the perspective provided by a joint appli- eyes as if to say, “I hate these idiots who do not cation design, it would have been difficult for credit know what they are doing.” to accept that credit checks were not essential to the overall process of making sales and processing or- Next he would ask how much horsepower I wanted. ders. Therefore, when determining information re- “How much can I get?” I would respond. He would quirements, all affected functions should be repre- say “five to 18,” as he rolled his eyes again. Then sented in the same room at the same time. he would ask such questions as, “Do you want wide tires, narrow tires, a rear bagger, a side bagger, manual start or electric start?” It turned out the trac- tor mower would cost between $800 and $2,800, Structured Interviewing depending on how it was configured. Ideally, I would not like to over-buy or under-buy. Once the The third mistake made in determining information salesperson realized I did not know what I was do- requirements is that the designers usually ask the ing, he would immediately start pushing the $2,800 wrong question: “What information do you need unit, suggesting I should go first class. Mowing my from the new system?” Though this is the obvious yard and first class are not two concepts I associate 8 Determining Executive Information Requirements with one another. Suspecting that I was being over- A straightforward, useful approach to interviewing sold, I would go to another dealership. Unfortu- executives (instead of simply saying “what infor- nately, I encountered the same experience at each mation do you need?”) to determine information dealership. requirements has been developed through research conducted at the MIS Research Center at the Uni- Finally, I went into a dealership where the salesper- versity of Minnesota (Wetherbe, 1988). The tech- son was not an order taker but a problem solver. He nique is based upon three different requirement/de- did not ask what features I wanted on the lawn termination methodologies, defined in Table 2. By mower. He asked other types of questions such as, combining questions from these three different meth- “How big is your yard?” My yard was about an acre. odologies, a comprehensive, reliable determination Next he asked, “How steep is your yard?” I told of conceptual information requirements can be him it had a gradual slope. He next asked, “What achieved. is the terrain like?” I told him it was natural, lumpy Before conducting the interview, an agreement of in spots. He wanted to know if I had fences or trees. the overall purpose of the business activity must be I replied, Yes, a fence and about 80 trees.” Next he established in a joint application design fashion. For wanted to know if I wanted my wife to use the trac- example, the objective of the order processing sys- tor mower. I said, “Are you kidding? This is her tem discussed earlier could be to ensure prompt, anniversary present!” correct delivery of orders to customers, maintain credit integrity, facilitate inventory management, and With that information, he walked over to a tractor ensure good shipment routing and scheduling. Once mower unit and said, “This is the one you need.” I this has been established, questions can be asked said, “How much?” He said, “Eighteen hundred that determine the information needed to ensure dollars.” Well, that was better than $2,800. I asked those objectives are accomplished. A basic model him why this particular unit. He said, “You have a for the information requirements interview is por- large yard so you want the widest blade. You need trayed in Figure 1. 12 horsepower to drive the widest blade, but you do not need 18 horsepower unless your yard is both The notion is to focus on issues that “back into” big and steep. You want wide tires to keep them information requirements. The specific questions from slipping into a rut, tilting the blade deck and asked are as follows. scalping the yard; you want a rear bagger so you can mow around the trees and mow by your fences one way one time and the other way the next time so you do not pack the grass; and you want electric Business Systems Planning (BSP) start because you have delusions that you are going (Source: IBM Corporation, 1984) to get your wife to use it!” Problem/Solution/Information Format Notice that what he did was extremely simple. He 1a. What are the major problems encountered in asked indirect questions that backed into my require- accomplishing the purposes of the organizational ments. He never specifically asked what features I unit you manage? wanted. For example, in an order processing system, prob- Using indirect questions is the creative skill of the lems include frequent stock-outs, allocation of lim- problem solver in sales versus the order taker. Prob- ited inventory to the wrong customers, and truck lem solvers creatively determine requirements departure scheduling. through less obvious, indirect questions. Those designing information systems need to do the same, 1b. What are good solutions to those problems? and executives should request that they do so. 9 Determining Executive Information Requirements Table 2: Comprehensive Interview Approaches, Implementations, and Developers Comprehensive Information System Approach Implementation Developer Specify problems and The executive interview portion of IBM decisions Business Systems Planning (BSP) Specify critical factors Critical Success Factors (CSF) Rockart Specify effectiveness Ends/Means Analysis (E/M Analysis) Wetherbe and Davis criteria for outputs and efficiency criteria for processes used to generate outputs Problems Solutions Information BSP Decisions Information Critical CSF Information Success Factors Ends Effectiveness Information E/M Means Efficiency Information Figure 1: Framework for Information Requirements Interview 10