Rattling SABRE—New Ways to Compete on Information

I have built my career, and American Airlines has built much of its business, around massive, centralized, proprietary computer systems. Developing these systems consumed millions of man-hours and billions of dollars, but their marketplace advantages were huge. As a result, our experience underscored the competitive and organizational potential of information technology. At the risk of sounding immodest, we helped define an era. That era is over. We are entering a new era, one in which the thinking that guided “best practice” as recently as five years ago is actually counterproductive. In this new era, information technology will be at once more pervasive and less potent—table stakes for competition, but no trump card for competitive success. As astute managers maneuver against rivals, they will focus less on being the first to build proprietary electronic tools than on being the best at using and improving generally available tools to enhance what their organizations already do well. Within their companies, they will focus less on developing stand-alone applications than on building electronic platforms that can transform their organizational structures and support new ways of making decisions. Who, by now, cannot recite the computer-based success stories of the 1970s and 1980s? SABRE, American Airlines’s reservation system, which eventually became a computerized reservation system (CRS), and Apollo, the other leading CRS, transformed marketing and distribution in the airline industry. American Hospital Supply’s ASAP order-entry and inventory-control system generated huge sales increases for the company’s medical products and turned it into an industry leader. United Service Automobile Association used its Automated Insurance Environment—a collection of telecommunication systems, databases, expert systems, and image-processing technologies—to consistently outperform its insurance industry rivals in service quality, premium growth, and profitability. Mrs. Fields Cookies relied on its Retail Operations Intelligence system, an automated store management network, to build and operate a nationwide chain of 400 retail outlets without a costly and stifling headquarters bureaucracy. These and a handful of other well-known computer systems (the Information Technology Hall of Fame, if you will) represent an important chapter in the application of electronic technologies to build competitive advantage and enhance organizational effectiveness. But it is time to turn the page. In 1984, F. Warren McFarlan published an influential article in HBR on the competitive potential of information technology.1 He asked managers to consider how information systems might benefit their companies. Could the technology build barriers to competitive entry? Could it increase switching costs for customers? Could it change the balance of power in supplier relationships? He went on to argue that for many companies the answer was yes. By being the first to develop proprietary systems, pioneers could revolutionize their industries. Increasingly, however, the answer is no. While it is more dangerous than ever to ignore the power of information technology, it is more dangerous still to believe that on its own, an information system can provide an enduring business advantage. The old models no longer apply. The Information Utility The new era is driven by the greatest upheaval in computer technology since the first wave of modern computer development 30 years ago. We are finally (and just barely) beginning to tap the real potential of computer functionality. As we change what computers can do, we must change what we do with computers. Think of it as the emergence of an “information utility.” Using superfast RISC architectures, hardware suppliers are delivering enormous processing power at remarkably low costs. UNIX and other software and communications standards are bringing unprecedented portability among different vendors’ products and among different classes of products. Software tools like relational databases, expert systems, and computer-aided software engineering are helping create powerful applications that meet specialized needs at reasonable costs. The ultimate impact of these and other technical developments is to give end-users greater power to shape their computer systems and manage their information needs. Increasingly, technology is allowing groups and individuals within companies to perform many of the functions once reserved for data processing professionals. It is hardly news to most managers that technology is changing faster than ever. Yet I wonder how many appreciate just how radical and rapid the changes are. Over the past two decades, price/performance ratios for computer technology improved at an annual compound rate of roughly 10 % . In recent years, those ratios improved at a compound rate closer to 40 % . This massive acceleration in performance will have profound implications for how computers are used and how useful computers are. Three features of the new environment will be particularly important. Powerful workstations will be a ubiquitous presence in offices and factories, and organizations will use them far more intensively and creatively than they do today. One of the paradoxes of the information age is that computers become easier to use as they become more powerful and complex. That’s what is so important about dramatic hardware advances like microprocessors with a million transistors on a chip. Personal workstations running at near supercomputer speeds will finally be powerful enough to be simple and thus truly useful. Meanwhile, new graphical user interfaces are creating screen environments (electronic desktops) that make it quicker for employees to become skilled with their workstations, to move between systems without extensive retraining, and to develop the confidence to push the functionality of their machines. In the not-so-distant future, computers will be as familiar a part of the business environment as telephones are today. They will also be as simple to use as telephones, or at least nearly so. As a result, companies will find it harder to differentiate themselves simply by automating faster than the competition. It will be easier for every organization to automate and to capture the efficiency benefits of information technology. This leaves plenty of room for competitive differentiation, but differentiation of a new and more difficult sort. Companies will be technology architects rather than systems builders, even for their most critical applications. The widespread adoption of standards and protocols in hardware, software, and telecommunications will dramatically recast the technology-management function. At American Airlines, for example, we have spent 30 years handcrafting computer systems. We like to think we’re better at this than most and that our skills in hardware evaluation, project management for software development, and systems integration have given us an important leg up on the competition. But we look forward to the day when we can buy more and more of our hardware and software from third-party vendors capable of tailoring their systems to our needs—and that day is rapidly approaching. InterAAct, our major new initiative for organizational computing, is a good example. Unlike SABRE, which incorporates a vast amount of AMR-developed technology, InterAAct is built around hardware and software provided by third-party vendors: workstations from AT&T, IBM, and Tandy; minicomputers from Hewlett-Packard; HP’s NewWave presentation software and Microsoft Windows; local area networks from Novell. We play a role in systems integration (in particular, merging the networks), but outside suppliers are capable of delivering more value than ever before. Of course, if we can buy critical hardware and software from outside vendors, so can our competitors. Our skills as electronic-tool builders, honed over decades, will become less and less decisive to our information technology strategy. This may sound like bad news, but we welcome it. We’re not in business to build computer systems; our job is to lead in applying technology to core business objectives. We don’t much worry if the competition also has access to the technology; we think we can be smarter in how we use it. Economies of scale will be more important than ever. We have entered the age of distributed computing, an age in which a young company like MIPS Computer Systems delivers a $ 5,000 workstation with processing speeds comparable to those of a $ 3 million IBM 3090 mainframe. Yet the amount of information required to solve important business problems also keeps growing, as does the capacity of telecommunications systems to transmit data quickly and reliably between distant locations. More than ever, then, the benefits of distributed computing will rely on access to vast amounts of data whose collection and storage will be managed on a centralized basis. The proliferation of desktop workstations will not erode the importance of scale economies in information processing. Consider the airline industry. American Airlines began working on a computerized reservation system in the late 1950s as the volume of reservations began to outrun our capacity to handle them with index cards and blackboards. In 1963, the year SABRE debuted, it processed data related to 85,000 phone calls, 40,000 confirmed reservations, and 20,000 ticket sales. Today there are 45 million fares in the database, with up to 40 million changes entered every month. During peak usage, SABRE handles nearly 2,000 messages per second and creates more than 500,000 passenger name records every day. As we enhance SABRE, we are aggressively replacing “dumb terminals” in travel agents’ offices, airline reservation offices, and airports with workstations capable of intensive local processing. But as a system, SABRE still works only in a centralized environment. The level of data collection and management it must perform dwarfs the demands of the 1960s just as thoroughly as the performance of today’s computers dwarfs the performance of their ancestors. The continued importance of scale economies has at least two major implications for information technology. First, truly useful computer systems are becoming too big and too expensive for any one company to build and own; joint ventures will become the rule rather than the exception. Second, organizations (like AMR) that have developed centralized systems will eagerly share access to, and sometimes control of, their systems. For companies to remain low-cost providers of information, they must tap the enormous capacities of their systems. Tapping that capacity requires opening the system to as many information suppliers as possible and offering it to as many information consumers as possible. From Systems to Information I do not mean to diminish the pivotal role of information technology in the future or to suggest that technology leadership will be less relevant to competitive success. Precisely because changes in information technology are becoming so rapid and unforgiving and the consequences of falling behind so irreversible, companies will either master and re-master the technology or die. Think of it as a technology treadmill: companies will have to run harder and harder just to stay in place. But that’s the point. Organizations that stay on the treadmill will be competing against others that have done the same thing. In this sense, the information utility will have a leveling effect. Developing an innovative new computer system will offer less decisive business advantages than before, and these advantages will be more fleeting and more expensive to maintain. The role of information technology has always been to help organizations solve critical business problems or deliver new services by collecting data, turning data into information, and turning information into knowledge quickly enough to reflect the time value of knowledge. For 30 years, much of our money and energy has focused on the first stage of the process—building hardware, software, and networks powerful enough to generate useful data. That challenge is close to being solved; we have gotten our arms around the data-gathering conundrum. The next stage, and the next arena for competitive differentiation, revolves around the intensification of analysis. Astute managers will shift their attention from systems to information. Think of the new challenge this way: In a competitive world where companies have access to the same data, who will excel at turning data into information and then analyzing the information quickly and intelligently enough to generate superior knowledge? On Wall Street, there are stock traders who wear special glasses that allow for three-dimensional representations of data on their screens. They need three dimensions to evaluate previously unimaginable quantities of information and elaborate computer models of stock patterns. Manufacturers Hanover has developed an expert system to help its foreign-currency traders navigate through volatile markets. In our industry, powerful new tools are helping us answer faster and more precisely questions we have struggled with for years. What is the best price to charge for each perishable commodity known as an airline seat? How do you reroute aircraft after a storm disrupts airport operations? How do you distribute your aircraft between airports? How do you meet the special needs of each passenger without pricing your basic service out of reach? As the process of analysis intensifies, decisions we once made monthly, we’ll make weekly. Those we made weekly, we’ll make daily. Those we made daily, we’ll make hourly. Consider yield management, the process of establishing different prices for seats on a flight and allocating seats to maximize revenues—that is, calculating the optimal revenue yield per seat, flight by flight. Yield management is certainly one of the most data-intensive aspects of the airline business. Computers review historical booking patterns to forecast demand for flights up to a year in advance of their departure, monitor bookings at regular intervals, compare our fares with competitors’ fares, and otherwise assist dozens of pricing analysts and operations researchers. During routine periods, the system loads 200,000 new industry fares a day. In a “fare war” environment, that figure is closer to 1.5 million fares per day. The initial challenge in yield management was to build software powerful enough to handle such demanding analyses. We spent millions of dollars developing SABRE’s yield-management software, and we consider it the best in the world. Indeed, we believe our pricing and seat-allocation decisions generate hundreds of millions of dollars of incremental annual revenue. For years, we guarded that software jealously. Since 1986, however, we have sold SABRE’s revenue-management expertise to any company that wanted to buy it. One of our subsidiaries—called AA Decision Technologies, many of whose members built our original yield-management applications—is knocking on the doors of airlines, railroads, and other potential customers. Why? Because we believe our analysts are better at using the software than anyone else in the world. Whatever “market power” we might enjoy by keeping our software and expertise to ourselves is not as great as the revenue we can generate by selling it. Similarly, Mrs. Fields has begun marketing to other retail chains the sophisticated networking and automation system with which it runs its cookie operations. Price Waterhouse is helping companies like Fox Photo evaluate and install the Retail Operations Intelligence system, the backbone of Mrs. Fields’s nationwide expansion. This is the competitive philosophy with which American Airlines is entering the new era: we want to compete on the use of electronic tools, not on their exclusive ownership. Computers and Competition: SABRE Reconsidered Perhaps no case study better illustrates the changing competitive role of computer technology than the evolution of the system that helped define the old era—SABRE. According to conventional wisdom on SABRE, the fact that American Airlines developed the world’s leading computerized reservation system generated substantial increases in traffic for us by creating market-power advantages over the competition. This has always been a difficult proposition to document. Analysts once pointed to so-called “screen bias” as a source of marketing advantage, even though the government-mandated elimination of such biases in 1984 produced no appreciable decline in bookings for American Airlines. Others argued that American’s access to CRS data regarding the booking patterns of travel agents gave us an incalculable information and marketing edge over our rivals—an argument that has proven groundless. Now the experts speak of a halo effect that by its very nature is impossible to identify or document.2 We are proud of what SABRE has achieved, and we recognize that it represents a billion-dollar asset to the corporation. But I have always felt the folklore surrounding SABRE far exceeded its actual business impact. SABRE’s real importance to American Airlines was that it prevented an erosion of market share. American began marketing SABRE to travel agents only after United pulled out of an industry consortium established to explore developing a shared reservation system to be financed and used by carriers and travel retailers. The way American was positioned as an airline—we had no hubs, our routes were regulated, and we were essentially a long-haul carrier—meant that we would have lost market share in a biased reservation system controlled by a competitor. SABRE was less important to us as a biased distribution channel than as a vehicle to force neutral and comprehensive displays into the travel agency market. My concerns about the conventional wisdom surrounding SABRE, however, go beyond the issue of market power. SABRE has evolved through four distinct stages over the past 30 years. In each stage, it has played different roles within American Airlines, and each role has had a different impact on the industry as a whole. Unfortunately, most analysts mistake the CRS distribution stage for the entire story. To do so is to invariably draw the wrong lessons. SABRE took shape in response to American’s inability to monitor our inventory of available seats manually and to attach passenger names to booked seats. So SABRE began as a relatively simple inventory-management tool, although by the standards of the early 1960s, it was a major technical achievement. Over the years, the system’s reach and functionality expanded greatly. By the mid–1970s, SABRE was much more than an inventory-control system. Its technology provided the base for generating flight plans for our aircraft, tracking spare parts, scheduling crews, and developing a range of decision-support systems for management. SABRE and its associated systems became the control center through which American Airlines functioned. American installed its first SABRE terminal in a travel agency in 1976, inaugurating its now familiar role as a travel-industry distribution mechanism. Over the decade that followed, we added new services to the database (hotels, rail, rental cars), built powerful new features to help travel agents offer better service, increased the installed base of SABRE terminals, and created a training and support infrastructure. SABRE now operates in more than 14,500 subscriber locations in 45 countries. Largely as a result of the proliferation of such systems, travel agents now account for more than 80 % of all passenger tickets as compared with less than 40 % in 1976. SABRE and its CRS rivals truly did transform the marketing and distribution of airline services. Today, however, SABRE is neither a proprietary competitive weapon for American Airlines nor a general distribution system for the airline industry. It is an electronic travel supermarket, a computerized middleman linking suppliers of travel and related services (including Broadway shows, packaged tours, currency rates) to retailers like travel agents and directly to customers like corporate travel departments. Speak with any of the 1,800 employees of the SABRE Travel Information Network, the system’s marketing arm, and you will hear that their division doesn’t treat American Airlines materially differently from the other 650 airlines whose schedules and fares are in the system. American pays SABRE the same booking fees as other airlines do. SABRE’s capacity to write tickets and issue boarding passes works similarly on other large carriers as it does on American flights. Although limited performance differences remain (largely as a result of SABRE’s technical heritage as an in-house reservation system), SABRE programmers are working to overcome these limitations and put all carriers on an equal footing in the long term. I don’t deny that there is some halo effect from SABRE that benefits American Airlines in the marketplace, although we have never been able to determine the magnitude or causation. But the core identifiable benefit American Airlines now receives from SABRE is the revenue it generates. This is not an inconsequential advantage, to be sure, but it is difficult to argue that the SABRE system tilts the competitive playing field in ways that uniquely benefit American Airlines. This is not necessarily how we would prefer it, but it is what the technology, the market, and the U.S. government demand. There is no compelling reason for a travel agency to accept a CRS that does not provide the most comprehensive and unbiased system for sorting through thousands of potential schedules and fares. If SABRE doesn’t do the job, another system will. SABRE’s industry-leading U.S. market share of 40 % means that rival systems account for three out of every five airline bookings. I receive weekly reports on our “conversion wars” with Covia, whose Apollo reservation system remains our chief competitor, and the other U.S.-based CRS systems. Subject to contract-term limitations that are established by the U.S. government, it takes only 30 days for a travel agent who is unhappy with SABRE to pull the system out and install a competing system. If a CRS can be replaced within a month by a rival system, can it really be considered a source of enduring competitive advantage? The old interpretations of SABRE simply no longer apply. As a group of HBR authors argued, “Early developers of single-source or biased sales channels should plan for the transition to unbiased electronic markets. That way, they can continue to derive revenues from the market-making activity.”3 The alternative, they might have added, is for the biased channel to disappear altogether in favor of unbiased markets offered by other suppliers. This is the future of electronic distribution. It is increasingly difficult, if not downright impossible, for computerized distribution systems to bind customers to products. Smart customers simply refuse to fall into commodity traps. (Indeed, American Hospital Supply has opened ASAP to products from rival companies.) It is increasingly difficult to design information systems that locked-out competitors or coalitions of locked-out competitors can’t eventually imitate or surpass. It is increasingly difficult for one company to marshal the financial resources to build new information systems on the necessary scale. We are applying these new rules outside the airline realm. AMRIS, a subsidiary of AMR, is developing a computerized-reservation and yield-management system for the hotel and rental car industries. Its power and sophistication will exceed anything currently available. We expect that the introduction of the Confirm system, scheduled for 1991, will affect pricing strategies and marketing techniques in the hotel and rental car industries in much the way Apollo and SABRE transformed the airline business. But we are not approaching the system itself in the same way we approached SABRE—at least three major differences stand out. For one, we are not going it alone. AMRIS has formed a joint venture with Marriott, Hilton, and Budget Rent-A-Car to develop and market the Confirm system. Moreover, there will be nothing biased about Confirm’s reservation functions—no tilted screen displays, no special features for the sponsors. Finally, the management aspects of the system, such as the yield-management software, will be generally available to any hotel or rental car company that wants to buy them. Confirm’s sponsors are participating in the creation of the most sophisticated software in the world for their industries; but the moment the system is operational, they will offer its tools to their competitors around the world. Not all companies will benefit equally from this new system. As is true with marketing or finance or employee development, some organizations will excel in manipulating, analyzing, and responding to the data Confirm generates. But no company will be locked out of access to the data or the opportunity to use it to compete. As in airlines and so many other industries, competition shifts from building tools that collect data to using generally available tools to turning data into information and information into knowledge. Building the Organizational Platform As with competition between companies, technological change will have profound consequences for the role of computers within companies. Until recently, I was not a champion of office automation. Work-stations were simply not powerful enough nor affordable enough nor easy enough to use nor capable enough of being integrated into networks to justify large investments in organizational computing. Indeed, a visitor to my office would be hardpressed to find more than a handful of personal computers on the desks of the information technology professionals. In the last few years, though, as a result of the technology changes I have outlined here, my caution has given way to genuine enthusiasm. But in this area too, it is time for new thinking. Understandably, given the earlier limitations of the technology, most companies approached office automation with an “applications” mind-set. They developed discrete systems to make administration more efficient, to improve planning and control, or to deliver particular services more effectively. We are taking a different approach. AMR has embarked on a multiyear, $ 150 million initiative to build an information technology platform modeled directly on the utility concept. This platform, called InterAAct, provides for the convergence of four critical technologies: data processing, office automation, personal computing, and networking. InterAAct will provide an intelligent workstation for every knowledge worker at AMR and will guarantee that every employee, no matter the rank or function, has easy access to a workstation. These workstations will be part of local area networks connecting work groups and a corporatewide network linking every location in the company, from departure gates in Boston to the underground SABRE facility in Tulsa, Oklahoma to the CEO’s office in Dallas/Fort Worth. The goal of InterAAct is not to develop stand-alone applications but to create a technology platform—an electronic nervous system—capable of supporting a vast array of applications, most of which we have not foreseen. InterAAct is an organizational resource that individuals and groups can use to build new systems and procedures to do their jobs smarter, better, and more creatively. It should eliminate bureaucratic obstacles and let people spend more time on real work—devising new ways to outmarket the competition, serve the customer better, and allocate resources more intelligently. InterAAct began to take shape in 1987, and rollout started last June. It will take at least three years to extend the platform throughout the AMR organization. We are approaching the project with four guiding principles. 1. The platform must give each employee access to the entire system through a single workstation that is exceptionally easy to use and that operates with a standard user interface throughout the company. 2. The platform must be comprehensive, connecting all managerial levels and computing centers within the company, and be connectable to other companies’ platforms. 3. The project must generate hard-dollar savings through productivity gains that are quantifiable in advance, and it should be rolled out in stages to ensure that it is delivering those hard-dollar savings. 4. The project must be managed as much as an organizational initiative as a technology initiative. Installing a powerful electronic platform without re-designing how work is performed and how decisions are made will not tap its true potential. Installing InterAAct is partly a matter of faith. But $ 150 million projects cannot be justified on faith alone. After extensive study (including in-depth analyses of how 300 AMR employees from different parts of the company actually spend their time), we estimated that extensive automation could produce enough hard-dollar savings to generate a 10 % return on the InterAAct investment. AMR’s standard hurdle rate is 15 % , so corporate directors with a pure financial mind-set would not have approved this project. That’s where faith comes in. We are confident that the “soft-dollar” benefits—better decisions, faster procedures, more effective customer service—will boost returns on InterAAct well above the hurdle rate. Still we are rolling out the project in stages and testing its impact along the way to be sure the hard-dollar savings materialize first. I don’t know how InterAAct will change our company’s organizational structure and work practices over the next five years. But I guarantee there will be major changes. Most large companies are organized to reflect how information flows inside them. As electronic technologies create new possibilities for extending and sharing access to information, they make possible new kinds of organizations. Big companies will enjoy the benefits of scale without the burdens of bureaucracy. Information technology will drive the transition from corporate hierarchies to networks. Companies will become collections of experts who form teams to solve specific business problems and then disband. Information technology will blur distinctions between centralization and decentralization; senior managers will be able to contribute expertise without exercising authority.4 We are currently at work on a series of InterAAct applications to reduce common sources of frustration and delay within AMR. Why should employees remain in the dark about the status of resource requests? On-line forms and electronic signature control, to be introduced later this year, will help speed such approval processes. Why should an employee’s personnel file remain locked away and inaccessible? A pilot project at the Dallas/Fort Worth airport allows baggage handlers to use a workstation to check how much overtime they have accrued. Eventually, employees should be able to file their own insurance claims or check on their reimbursement status. With respect to bureaucratic procedures, the potential of an electronic platform is obvious: eliminate paper, slash layers, speed decisions, simplify the information flows. Other organizational possibilities are even more far-reaching. InterAAct standardizes spreadsheets and databases, provides direct access to the corporate mainframes, and will eventually support automatic report generation. The new ease and speed with which analysts will be able to accumulate and disaggregate data, conduct “what if” scenarios, and share information should accelerate the planning and budgeting process. It’s not our job to design a new planning process. InterAAct gives our analysts the potential to redesign systems to best suit their needs. Finally, and perhaps of greatest importance, InterAAct will allow senior executives to make their presence felt more deeply without requiring more day-to-day control. Eventually, executives should be able to practice selective intervention. The information system, by virtue of its comprehensiveness, will alert senior managers to pockets of excellence or trouble and allow them to take appropriate action more quickly. Over time, the role of management will change from overseeing and control to resolving important problems and transferring best practices throughout the organization. Who Needs the CIO? The ultimate impact of the hardware, software, and organizational developments I have described is to proliferate and decentralize technology throughout the organization. Piece by piece and brick by brick, we and others are building a corporate information infrastructure that will touch every job and change relationships between jobs. Much work remains to be done. We need better tools, more connectivity, and richer data that reflect the real business needs of our companies. But in all these areas, momentum is moving in the right direction. As technology reshapes the nature of work and redefines organizational structures, technology itself will recede into the strategic background. Eventually—and we are far from this time—information systems will be thought of more like electricity or the telephone network than as a decisive source of organizational advantage. In this world, a company trumpeting the appointment of a new chief information officer will seem as anachronistic as a company today naming a new vice president for water and gas. People like me will have succeeded when we have worked ourselves out of our jobs. Only then will our organizations be capable of embracing the true promise of information technology. 1. F. Warren McFarlan, “Information Technology Changes the Way You Compete,” HBR May–June 1984, p. 98. 2. For a comprehensive review of CRS technology in the airline industry, see Duncan G. Copeland and James L. McKenney, “Airline Reservation Systems: Lessons From History,” MIS Quarterly, June 1988. 3. Thomas W. Malone, JoAnne Yates, and Robert I. Benjamin, “The Logic of Electronic Markets,” HBR May–June 1989, p. 168. 4. For a good overview of the organizational possibilities, see Lynda M. Applegate, James I. Cash, Jr., and D. Quinn Mills, “Information Technology and Tomorrow’s Manager,” HBR November–December 1988, p. 128.

Rattling SABRE—New Ways to Compete on Information

Partner Center