Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Productivity Through Automation John ]. Alexancler, Jr. * Automation is widely viewed as a key way to increase produc- tivity. At Reynolds Metals Company the use of automation to improve productivity is the focus of the company's planning and development efforts for the 1980s. Its specific objective is to use automation to clouble the productivity of its salaried work force by 1990. A brief profile of the company will give some idea of the extent of this effort. Reynolds is a Fortune 100 producer of primary and recycled aluminum. Operations are conducted by 12 divisions that ex- plore, mine, ship, and refine bauxite; reduce alumina to primary aluminum; recycle scrap aluminum; and fabricate aluminum into a wide range of products. An international division manufactures and sells overseas. About one-third of our 30,000 employees are salaries! and work in 50 plants, 25 sales offices, and the corporate headquarters in Richmond, Virginia. Management, of course, is the key word in such an effort. And just like the management of any other corporate function, manag- ing information processing is itself a process. Specific tasks re- quired to carry out the process can be identified. In addition to the conventional maxim "plan, organize, and control," ~ would add "strategize, rationalize, and market." *John J. Alexander, Jr. is senior vice-president of management information sys- tems, MCI Telecommunications Corporation, Washington, D.C.; and former corpo- rate director of administration, Reynolds Metals Company. 101
102 AL4NAG~G MICROCOMPUTERS Reynolds' planning process for automation was developed dur- ing 1978 ant! 1979. The planning cycle begins in the spring when an update and an automation planning manual is issued, telling all division and plant automation managers what they need to do to develop an automation plan for their unit. Corporate manage- ment provides some structure and asks managers to cletai} their plans and programs. The cleadline for responses is October and coincides with the completion of the company's annual business planning cycle. Reynolds encourages but floes not require interaction between business and automation planning. The analysis of the results of one cycle consumes the October-to-May period and provides in- put and structure for the next cycle. Thus, planning is a continu- ous process. The first five planning cycles are summarized below. · 1979-80 patting cycle. In this first planning cycle har(l- ware was standardized and ciata was identifier! as an issue. A study was commissioner! that provicled an understanding of logi- cal and physical views of data and the tools for analysis. · 1980-81 planning cycle. The corporate staff proposed a net- work, which has been installed, and the salaried employee was identifier! as a target for major productivity improvement. · 1981-82 pawning cycle. The System/38 was added to the approved hardware list, experimentation with personal com- puters was encouraged, and IBM's business systems planning (BSP3 technology was introduced. · 1982-83 pearling cycle. The IBM Displaywriter and per- sonal computer were adcled to the approved hardware line, and a study of information architecture was begun. · 1983-84 piaurling cycle. The need for additional central computer capacity to keep balance with the growth of distributed processing was determined. From the early planning cycles we concluded that although the annual planning process is essential, looking at information re- quirements for the short term is not sufficient. We had to take a much longer look, and that required a strategic goal. In 1980 we articulated such a Tonger-term planning goal. This goal was to provide all salaried employees with the auto- mation tools they need to manage the information necessary to perform their jobs and double their productivity by 1990. In de
PRODUCTIVITY THROUGH AUTOAIATION 103 fining this goal we also described the rationale behind it, pro- jected the rate at which it couict be achieved, identified the major roadblocks to accomplishing it, developed a marketing approach to delivering the necessary automation tools, ant! livened the king! of changes in the organizational culture that would have to occur to achieve the goal. Although this is a very ambitious goal, the company believes it must be accomplished if Reynolds is to be productive and profit- able over the next clecade. Seeking a rationale, we looked at what our salaried employees do ant! recognized that information plays a very large role in their jobs. They receive information in the form of memos or reports and store some of it, mostly in five-drawer file cabinets. They use salaries! secretaries to retrieve it or go after it themselves, look at it, manipulate it, reformat it, analyze it, and send it on to somebody else. Automation can provide some sup- port for every one of these information processing operations. It is not going to reduce the human factor to zero, but it can increase procluctivity. Based on our projections about rates of change in cost of auto- mation components and people, we projected that 20 percent per year reduction in the cost of automation components is sustain- able for this decade. In the past five years, people have increased in cost an average of 10 percent per year at ReynoIcis. In some areas we have carefully tracked what a work group was doing before automation was applied and what was required to accomplish the same tasks after automation was applied. In these cases we can demonstrate improvements in productivity ranging from 100 to over 500 percent. The major roadblock to automation lies in taking the conven- tional approach to developing automation solutions to business problems. This approach calls for custom-designed systems to solve the information requirements of an inclividual or a group of individuals. At Reynolds what we call a system typically takes about three years to develop and directly affects about 15 people. With a limited set of resources, 150 people at Reynolds build and maintain systems. Half maintain the old systems and half build new systems. Using the 1980 work force as a base we assumed that the 1,000 individuals then using automation were representative of the to- tal. Extrapolating the systems work clone from 1977 through 1979 to the entire work force yielded! a backlog of approximately
104 MANAGING MICROCOMPUTERS three decades needed to automate all 11,000 salaried employees. We concluded that we needed a new approach. The alternative we developed is a marketing approach to pro- viding information services within a major corporation. Market- ing provides the answers to three critical questions: Who is our customer? What is our product? How is the product distributed? We began by recognizing that our "customer" is the individual employee who uses a computer or terminal. We classified these individuals by need and identified common information-handling requirements. This allowed us to organize the market of employ- ees into a handful of segments and to develop delivery vehicles for automation services to reach each segment. When analyzing such a market, it is helpful to consider whether it consists of all employees; various segments of employees like managers, professionals, clerks and secretaries, and other indi- viduals defined as office workers; or all salaries! employees. Rey- nolds settled on the last category as the most appropriate for its automation program. If the salaried employee is the customer, what is the product? I believe there are only two broad categories of automation ser- vices: custom systems and standard vehicles. Information professionals have been building custom data processing systems for 30 years. Each one has been a specific so- Jution to a functional or individual need, with all the attendant problems of limited life, costly maintenance, and general dissatis- faction with the disparities between what the customer wanted and what he or she got. An alternative to custom-designed systems is a standarcI ve- hicle, which is a general solution to an information problem pre- sented in a sufficiently friendly fashion for the individual to as- sume personal responsibility for it. The spreadsheet approach to presenting and analyzing financial data as embodies! in Visicalc is a classic example of a standard vehicle. Of course, standard ve- hicles have their own problems of documentation, standardiza- tion, and control. To get a custom solution a user specifies requirements to an information system analyst, who proceeds to build a custom prod- uct. This can be called systems development. By planning for the conception, birth, growth, maturity, and death of the product over a life cycle, systems development explicitly recognizes the obsolescence of the custom solution. By contrast, with a standard
PRODUCTIVITY THROUGH AUTOMATION 105 vehicle the user is educated into a solution without having to mas- ter the mysteries of electronic data processing. The educating is done by a professional who becomes the coach, informs the user of corporate stanciards, and helps select the best technology to sat- isfy the user's information needs. At Reynolds we started with the assumption that eventually every salaried employee will have a terminal. Every terminal will be tied to a network, and that network can access any computer. Although this is not something we can afford to do today, at least we have defined the problem, we know how to solve it, and it will become cheap enough to implement in the future. It may cost twice as much today to deliver a given service automatically as it does to deliver it manually, but if our cost projections hold true, crossover occurs in two years, after which automation is less expensive. Users will bear the cost of the terminal and a pro rata cost for using the network. We assume they will justify those devices based on productivity improvements of the staff. Other possible justifications include better service or new kinds of service that could not be providecI in any other fashion. But the focus is on productivity. To determine what kinds of automation wouict be needed we began by identifying six delivery vehicles for automated informa- tion processing: wore! processing, electronic mail, data inquiry, transaction processing, technical computation, ant! business analysis. We then took the personnel system's function cocles for all salaried employees and mapped them against the six delivery vehicles to identify the size of the market for each vehicle. Secretaries can obviously use word processing, but the market segment for this vehicle is much larger. Lawyers are wordsmiths; programmers and systems analysts are in the business of lan- guage translation; ant! public relations and purchasing depart- ments already have word processing capabilities. We currently have over 400 terminals providing word processing support. Es- sentially, what they do is format and edit words, but these capa- bilities can easily be expancled. Add technical computation and business analysis, and not just words but an entire report can be formated. Wrap an electronic envelope around a memo or a report, plug it into electronic mail, and it can be delivered to the recipient. Only managers are considered an appropriate market for elec- tronic mail. Among the things managers do is send and receive a
106 A~1NAGING MICROCOMPUTERS lot of memos. Electronic mail appears to be a fine vehicle for send- ing, receiving, and following up on memos. Currently Reynolds has over two dozen terminals that are Gentling and receiving memos by electronic mail. This includes a few intended primarily for international communications. Automatic follow-up, schedul- ing, ant! calendaring routines are available. Data inquiry services are "marketed" to employees who need access to public information or large data banks. In 1980 we had only a handful of terminals justifier! for this purpose. To deter- mine who neecled what kinds of data we brought in a ciata dic- tionary, which allows users to define the information they need to access. To meet these needs we have bought two language- processing systems: INQUIRE ant! INTELLECT. INQUIRE is a text-processing language with very powerful Boolean logic capabilities. Our research librarians finc3 it an effec- tive tool for abstracting, storing, and retrieving information about company reports and projects. The same language is being developed for use by the legal department to store abstracts of legal contracts and maintain an inclex of where those contracts are physically stored. INTELLECT is a parsing program that analyzes an English sentence types! on a computer terminal and generates a call to a specific database. The first application of INTELLECT has been in support of our Human Resource Development (HRD) search activity. In addition to database and query-language facilities, the Sys- tem/38, with its own capabilities, is providing outstanding ser- vice and accounts for the largest segment of our recent growth in the area of data inquiry. Further, we are just becoming comfort- able with the database tool on the personal computer. ~ expect that as we become proficient in defining ciata for these kinds of systems anyone who has a terminal will be able to use data inquiry to get access to needed information. Data inquiry will largely supplant printed reports as the vehicle for distribut- ing information from automated systems. Transaction processing is our largest delivery vehicle. In three years this application has grown 50 percent to over 900 terminals that support clerks who enter and process transactions. All termi- nals now exercise systems built to process transactions in a con- ventional manner. We have defined a requirement to handle the other 5,000 forms that the company is currently processing man
PRODUCTIVITY THROUGH AUTOMATION 107 u ally. We are now seeking a universal form processor that will allow a user to call up on a terminal any form used in the company, fill in the blanks, and send it by electronic mail to the individual or computer that has to review it, approve it, or process it. If this capability is acquired, we believe it wouIct be used by almost everyone. Technical computation is used by our technicians, engineers, and scientists. Timesharing has been the delivery vehicle, ant! the applications are mainly specializecl routines the company has pur- chased or the users have written for themselves. Business analysis was the final delivery vehicle we identified. In many companies business analysts, planners, accountants, and others have long used automation as a vehicle for improving their productivity. At the beginning of 1980, however, Reynolds had no terminals to support such individuals. This situation has changed with the availability of color graphics and analytic too! packages. A spreadsheet computer program such as Visicalc, which runs on a personal computer, is very popular; we wish it were available now on larger IBM machines. When it is we believe the number of terminals used by business analysts will grow even more rapidly. How is the investment in various automation delivery vehicles justified? I have already explained that in orcler to determine the market for each of the six clelivery vehicles we mapped the num- ber of people in each functional job code against their potential to use a primary delivery vehicle. The resulting match-up was also used for justification purposes. Obviously, we expect most users to employ more than one cleliv- ery vehicle. For example, secretaries who primarily use word pro- cessing also process transactions and inquire into databases. But the justification for this investment requires an improvement in productivity that is usually assumed to be measurable only by the primary focus. Once a terminal is installed for primary access, of course, other clelivery vehicles can be made available. When we matched job functions with delivery vehicles we found that terminals for about 57 percent of the salaries! employ- ees would be justified by transaction processing. The second larg- est justified use (15 percent) was for technical computation. This is because Reynolds is a manufacturing and technology company and has a significant cacire of engineers and scientists. Three per- cent of the staff was identified as managers, and shouict have cle
108 MANAGING MICROCOMPUTERS vices justified for electronic mail. The balance was about evenly split between word processing (10 percent), database (9 percent), ant! business analysis (6 percent). Quite a different picture emerged when we looked at the actual use of delivery vehicles compared to the justified (or projected) use. (See Figure 2.) The productivity improvement in one primary area had provided the justification, but microcomputers are in fact blurring the lines between vehicles, causing changes in job function and rebalancing the workload. What seems to be happen- ing is that automation is changing the jobs of salaried employees. The individual delivery vehicles have had widely varying growth rates between 1980 and 1983. Transaction processing, which began with the largest base, has been growing steadily at about 15 percent per year. Technical computation doubled in 1980 and grew another 60 percent during 1981. Word processing more than doubled in each of its first three years and is approaching 50 percent penetration of the market. In all, these three vehicles should satisfy over 80 percent of Reynolds's automation needs when fully implemented. Data inquiry, business analysis, and electronic mad! began the clecade at ground zero. In the last two years, both data inquiry and business analysis have broken out of the pilot/test modes and have achieved market acceptance. Electronic mail has not been as successful, and research is now uncler way to find out why. With its market reasonably defined and delivery vehicles in place, Reynolds adciressed the issue of the rates of growth we could manage while meeting our encI-of-the-decade obj ective. As a percentage of our salaried work force, inclivic3~uals using automa- tion grew from a base of 6 percent at the start of 1979 to 21 per- cent in January 1983. A compound growth rate of 30 percent per year from 1983 on would achieve the 1990 objective of 100 percent automation. Actual growth has been over 33 percent per year, compoundecI. As productivity improvements are achiever! the sal- aried work force needed to handle a given volume of business is being recluced. This makes 100 percent automation delivery by 1990 even more likely. Along with the marketing and growth aspects of its long-term plan, Reynolds has identified a number of organizational issues that demand attention. These can be broadly diviclec3 into technol- ogy, management, ant! cultural issues. Computers are available in a dazzling variety of sizes, capabili
PRODUCTIVITY THROUGH AUTOMATION JUSTI Fl ED USE ACTUA L US E / >W - 57% Transaction Processi ng 28% Transaction Processing / 10% Technical Computation 8% Business Analysis 10% Word Processing ,~3% Electronic Mail //\_ 9% Data Inquiry // ~ _ 6% Business Analysis ~ 15% Technical / Computation X / 20% Data Inquiry 20% Word A/ Processing -14% E lectron ic Mail 109 FIGURE 2 Justified versus actual use of automation delivery vehicles by Reynolds Metals salaried employees.
110 AL1NAGING MICROCOMPUTERS ties, and costs. To deal with this diverse technology Reynolds simplified its perception to three sizes: large, medium, and small. Two large AmciahIs manage the corporate information systems and interconnect all other machines. Medium-sizec! IBM System/ 38s manage plant ant! department information needs. The IBM Displaywriter and personal computers provide indiviclual auto- mation capability. All of these machines are connected through a network. Although some form of each delivery vehicle capability is currently available on all user devices, some are better than others. The long-term objective is for all user devices to provide high levels of all six automation functions. To successfully change the mode of distributed processing in a company (that is, to change from manual to automated process- ing throughout), the three key variables of hardware, software, and management must be organized in some way. The question is how? If we simplify the possible operating environment for each of these variables to either totally centralized or totally de- centralized, a number of combinations are possible. These can be viewed as the eight corners of a cube (Figure 31. For example, centralized electronic data processing (EDP) func- tions best when all three variables (hardware, software, and man- agement) are centralized. At the opposite extreme is the personal computer, with all three variables totally decentralized. At the back plane of the cube in Figure 3, where management is centralized, three combinations are possible. If only software is relaxed (decentralized), development is distributed. Alterna- tively, relaxing hardware leads to distributed processing. Decen- tralizing both hardware and software with centralized manage- ment requires strong efforts toward standardization. The face plane represents clecentraTizec3 management. If this occurs when hardware and software are both centralized, frustra- tion is the usual result. When hardware remains centralizer! and software is relaxed, timesharing and remote job entry are the result. Centralizing hardware and relaxing software are equiva- lent to purchasing packages. Most organizations consist of some mix of all or most of the eight extremes of automated processing. The important point is for organizations to know where they stand and where they are likely to be in the future. The culture of an organization poses a final set of concerns for automation efforts. Although it is unplottec! in Figure 3, a culture
PRODUCTIVITY THROUGH AUTOA~1TION D I ~ 1 Or 1§ 1 C) 1 ~ 1{ Distributed -~' Processing / I / ! Pu rchased _ Packages Central ized EDP / Frustration ~< $~// D - ~ ' ) Aft__ __ _ _ Remote Job Entry / :7~) ~ Standardization Personal Computers ~ ~ //A ~ r Distri bused Development C = Centralized D= Decentralized FIGURE 3 Operating space of automated information processing. is present in every company. The culture of an organization must change as it moves through the hardware-software-management cube. In general, I believe that automation professionals neec! to reorient themselves from custom systems to standarc! solutions. This frees key personnel to become coaches and focuses a smaller cadre on technology planning, evaluation, and selection. Other employees need to accept automation education, devices, and net- works that give access to information. This allows indivicluals to accept responsibility for their own information requirements. A comparison of the cultural implications of automation with those of human resources clarifies the management issue. Along with money and machines, people and information are resources every organization must manage. The personnel function is recog- nized by most organizations as a key one, and reports are made to the chief executive officer or one step below. Well-run personnel 111 D
112 MANAGING MICROCOMPUTERS departments provide expertise in recruiting, compensation, de- velopment, evaluation, and training. However, management of the human resources is not the responsibility of the personnel de- partment but of managers at various department and section levels. Information management should be considered in the same light. For too long it has been clelegatect to the EDP or MIS de- partment. Yet computers are no more complex to manage than people. Every manager and information worker needs to under- stanc! what information is needed, where it is, who controls it, how good it is, and how to get it. Plans to improve information avail- ability, cut costs, and measure performance should be made annu- ally, just like human resource clevelopment plans. How to make people recognize that information management is part of every manager's job is another question. Leadership needs to come from the information systems ranks. Identifying the chairman, president, or business unit vice-president who will be the role mode] in the organization is essential. Articles in periodi- cals such as the Harvard Business Review, Fortune, and Busi- ness Week, which popularize the role of information management, will also be helpful. Reynoicis has given priority to four technical issues. First, we need a data dictionary that spans the full range of automation devices from personal computers to departmental machines to corporate mainframes. The second issue is access. Once we know what information is located where, access must be available to all who need it. The access issue leads directly to privacy ant! secu- rity issues. Information must be available only to those who shouts! have it, and in a fashion that assures individual privacy. The fourth technical issue involves optimal selection of devices and networks. Presently we analyze stated needs and select the best devices to meet these needs. We then build networks to con nect the crevices. A better approach would be to feet! all sources and uses of data to a mode! that provides the best combination and use of computers, terminals, and networks. On the management side of the equation, we have identified a number of difficult open issues. One of the major issues we will face, both as individuals and as organizations, is how to imple- ment productivity improvements. If a group of 10 clerks in- creases its productivity by 50 percent, a fairly straightforward
PRODUCTIVITY THROUGH AUTOMATION 113 management response would be to allow turnover or reduction in the work force. However, if a marketing manager or an engineer- ing manager is given a device that cloubles productivity, it is much more clifficult to clecicle how to take advantage of it. Do you give the marketing manager two markets to manage? Do you give the engineering manager both engineering and marketing to manage? It will be a major challenge for the management of Reynolds and other companies to fine! ways to restructure jobs to take acl- vantage of the improvements in productivity made possible by automation. Inevitably, some people will be displaced. Managers in our company and the country at large are going to have to ad- dress the issue of retraining these people. A second management issue we are now facing in the course of increasing security for corporate data is the question of who owns what data. ReynoIcis has bought a sophisticated computer pro- gram that requires identifying who owns each piece of data, who can access it, who can change it, and who can clelete it. This raises some interesting questions. For example, does a piece of tax infor- mation in a division system belong to the tax department or to the division? A third issue involves measuring productivity. Economic feasi- bility has been blithely assumed, but it needs to be tested at every point. On the one hancI, Reynolds has established a goal to double the productivity of its salaries! work force. On the other, we don't even measure the productivity of our salaries! work force today. We can measure the ratio of the total number of salaried employ- ees to pounds of aluminum or dollars of sales. On an individual basis, however, we don't know what productivity is and we don't know how to measure it. I believe this question of how to measure individual productivity is an issue we will wrestle with at least for the rest of the decacle. A final management issue involves multiple functions and re- sponsibilities. In the past auditors have taken a lot of comfort in the careful separation of responsibility. One person requisitions a purchase, a second approves, a third receives, ant! a fourth pays the invoice. Automation makes it possible for a single individual to perform more than one function. Reynold's management, as well as its auditors, must consider what kinds of controls to build into automated information processing systems. These controls /
114 MANAGING MICROCOMPUTERS must provide adequate protection for corporate assets, assure proper processing of financial transactions, and at the same time achieve the necessary improvements in productivity. I think the broad conclusions to be drawn from Reynolds' ef- forts are clear and important. First, information processing is just too important a component of every salaried employee's job to subcontract it to information specialists or anyone else. It needs to be a part of each employee's job clescription. Second, the decreasing cost of automation technology and the increasing cost of people will not only permit but will mandate that every salaried employee's productivity be improved through the use of automated technology cluring the course of this decade. Thirst, redesigning jobs to take advantage of the productivity potential provided by automation is going to present major chal- lenges to management in this decade. Microcomputers are machines. Like all machines they have ad- vantages and disadvantages. To deal with them successfully or- ganizations need an automation strategy, an understanding of how microcomputers fit, an automation plan, and a mechanism to monitor performance. ReynoIcis is only one company that has begun this process. I believe that all firms that successfully automate their infor- mation functions will develop a competitive ecige over those that clo not.
