Flying by the Seat of a DSS
American Airlines is one of the largest airlines in the United States, with the 1999 revenues of AMR Corp, its parent company, approaching $20 billion. American, headquartered in Forth Worth, Texas, has long been considered a leader in the use of information technology, gaining that reputation with its development of SABRE, its computerized airline reservation system, in the early 1960s. SABRE gave American a clear strategic advantage for a number of years, and over time it came to be used by other airlines as well as by travel agents (providing American with additional income). However, American Airlines has not been a company that sits on its IT laurels. The group that developed SABRE became The Sabre group, a separate company from American Airlines but a subsidiary of AMR Corp., and sabre has continued to do the systems development for American Airlines. Over the years Sabre enhanced and improved the SABRE system while also developing numerous other new, critical systems for American, such as its pricing, scheduling and yield management systems. It also developed systems for other airlines and even for nonairline transportation companies.
During the 1980s the competitive environment of the airline industry changed drastically because the U.S. government deregulated the industry by removing most nonsafety related regulations governing airline routes and flights. Suddenly, for the first time in decades, airlines were free to make their own competitive route and scheduling decisions. Existing airlines quickly sought to expand into routes in territories dominated by their competitors. They also opened new routes and abandoned existing unprofitable routes that previously had been government mandated. Start-up airlines appeared and identified all kinds of opportunities. They opened new routes where they believed a potential demand existed. They began service on existing routes in an attempt to steal some of the business of older airlines, often competing through lower prices based on diminished services. Thus, airline deregulation resulted in a fiercely competitive environment for all the existing airlines. American’s management team realized that it must become more nimble in order to survive. To do that it had to be able to forecast more quickly and accurately the financial impact of any company decision to add or subtract flights from its schedule. It also needed to be able to forecast the impact on American of other carriers’ decisions to add or change routes that are in competition with its own service.
In this new competitive environment, several large airlines did fail. Pan American Airways, the great pioneer in overseas passenger flights, was the most conspicuous example of an airline that finally had to shut down. This development made it clear to the American Airlines management team, including CEO and chair Bob Crandall, that accurate and speedy forecasting was urgent if the company wished to remain successful. Unfortunately, forecasting systems in those days consisted only of PC spreadsheets and calculators. Using low-level technology, American analysts had to work for days merely to estimate the effects of a change in only one single flight. Forecasting the financial effects of a large schedule change was well beyond the abilities of American’s planning department. American soon moved to develop a powerful mainframe-based forecasting system that is named Integrated Forecasting System (IFS).The 1989 collapse of another major competitor, Eastern Airlines, gave American the opportunity to test its new system. During those years airlines were very busy building hub systems, and to ward the end of the decade, American had decided it needed to build a new hub somewhere in the southeastern United States. Atlanta, Georgia, emerged as the leading candidate for the location before the Eastern bankruptcy became public. However, the Eastern collapse opened up another possibility because Eastern Airlines had dominated the profitable New York-Florida routes. Now American had to consider Miami, Florida, as another possible hub location. Management at American quickly turned to its newly completed IFS for supporting data to help the management team make a decision as to which would be the besr site for their new hub. The problem for the IFS was that this decision was infinitely more complex than that of making a change in the schedule of a single flight. A new hub meant a dramatic change in many flights and routes, and the new IFS would have to forecast the results of all these massive changes. Moreover, it would have to it for two separate locations. The planning department put IFS to its first big test, and it passed with flying colors.
To complete the whole task IFS only had to work a total of eight hours। The system output indicated that Miami would be more advantageous financially, and Crandall quickly presented that plan to his board of directors who then approved it. According to Brad Jensen, vice president of application development for Sabre Technology Solutions, the Miami hub has generated hundreds of millions of dollars in profit in the years since that decision
The new IFS system did have a major problem-it had to handle am immense amount of data. The system draws on 60 source data files, supplying both internal and public data. The files provide such varied information as passenger demand, fares, aircraft seating capacities, unit costs per flying hour, and even competitor flight schedules. To try to understand the quantity of data, consider that American has about 4,000 flights daily, and that the company must track at one time approximately 400,000 flight combinations (that is, pairs of cities that are or might be linked by flights). Moreover, the success of IFS on the hub questions generated another problem for American. The demand for the use of the system rose rapidly. However, users were unwilling to wait eight hours to get a response to their questions. In order to speedily handle such an enormous amount of data while responding to the greatly increased number of requests, The Sabre team decided the system would have to be moved to a parallel-processing architecture. A project to accomplish this goal was estimated to cost an immense $16 million. Fortunately, Crandall supported the project because the SABRE reservation system had been so successful. In fact, according to reports, when Crandall first was told of the proposal, including the cost, he jumped out of his chair yelling, ‘’I want the system.’’ Sabre management also was strongly in support of the project. Thomas M. Cook, president of Sabre Technology Solutions, remembers that, ‘’There was a realization among the senior management at both companies that in order to have significant breakthroughs, you have to take significant risk.’’
The development team made two critical decisions in 1991. The first was to convert to a client/server platform that would not only provide a parallel-processing environment but would place control of the upgraded system on the desktops of the users. Thus, the users would be able to initiate and manage their own forecasts right from their own PCs. The second decision was to adopt a new and, at that time, relatively unproven technology, object oriented programming. On reflection a third decision has been seen as critical to the success of the project. The development team decided it must work closely with the end users, American’s capacity planning department, and so the team was housed on the same floor as the planners. The result was a very close working relationship between the two sides and a successful project. Adding to the project’s success was the commitment by senior American executives to attend weekly project meetings, thereby lending not only their strong and visible support to the project but their design and usage insights as well.
The project was completed in 1996, and American found itself with a much faster and more powerful IFS. Sabre continued to improve the system so that the queries it can handle today are much more complex even than were those in 1996. It now forecasts within 3 minutes the company’s market share for its 4 million services worldwide as well as 400,000 city pairs. With the new client/server system, planners can actually get answers to hundreds of questions about a potential single flight change, whereas, previously, the mainframe IFS could only handle two or three questions. American planners are now able to do much more planning, do it more effectively, and yet do it with a reduced staff. American’s planners use IFS constantly to find ways to cut costs and increase revenues. The planners analyze the financial impact of adding or deleting flights, adding service to a new city or dropping an existing service, or changing the size of an aircraft for a specific flight or route. They even use it to project the financial impact of schedule changes made by American’s competitors. One interesting example of a result of IFS projections was American’s decision to replace some jet planes with propeller planes. Despite the fear of many that the change would result in a loss of passengers, the system showed that American could redeploy the jets by adding flights to longer, more profitable routes. American’s cost-per-mile declined while profits rose. All-in all American planners and management have gained a much deeper and more complete understanding of its customers, its competition, and the market place. Moreover, management claims American’s profit have gone up about $500 million as a result of the first 2 years of using IFS.
Although American Airlines has benefited greatly from AMR’s ownership of Sabre, that relationship is about to change. Over the years Sabre has developed many systems for other companies, including selling versions of IFS to the London subway system and to the French railways. However, Sabre’s close ties to American Airlines have been seen by other airlines as a problem. According to James Poage, senior vice president of worldwide marketing for Sabre, the problem ‘’we faced in the past is [other airlines would] say ‘Because of your relationship with American Airlines, is our information secure?’ ‘’ Because of this fear Sabre garnered no significant airline outsourcing contracts during 1988 and 1999. As a result, and with the goal of increasing the potential for Sabre, AMR announced on December14, 1999, that it is spinning Sabre off, making it a fully independent technology company. Sabre president and CEO Williams J. Hannigan explained the expected result, saying ‘’with our independence, it’s more clear that success for Sabre won’t mean enrichment for American. That’s real important to those carriers. Soon AMR will be able to sell its systems to any company and will be free to develop for any airline.’’
Sources: Scott McCartney, ‘’Airlines Find a Bag of High Tech Tricks to keep Income Aloft,’’ The wall Street Journal, January 20, 2000; Stacy Collett, ‘’Update: Sabre to Sell Software Once Controlled by American Airlines, ‘’ Computerworld, December 14, 1999;Polly Schneider, ‘’clear Skies Forecast,’’ C10 Magazine, February 1, 1999; Lauren Gibbons Paul, ‘’From Harrowing to Heroism,’’ C10 Magazine, September 1, 1998; and Linda Rosencrance,’’ AMR to spin off Sabre, ‘’ Computer world, December 12, 1999.
CASE STUDY QUESTIONS
During the 1980s the competitive environment of the airline industry changed drastically because the U.S. government deregulated the industry by removing most nonsafety related regulations governing airline routes and flights. Suddenly, for the first time in decades, airlines were free to make their own competitive route and scheduling decisions. Existing airlines quickly sought to expand into routes in territories dominated by their competitors. They also opened new routes and abandoned existing unprofitable routes that previously had been government mandated. Start-up airlines appeared and identified all kinds of opportunities. They opened new routes where they believed a potential demand existed. They began service on existing routes in an attempt to steal some of the business of older airlines, often competing through lower prices based on diminished services. Thus, airline deregulation resulted in a fiercely competitive environment for all the existing airlines. American’s management team realized that it must become more nimble in order to survive. To do that it had to be able to forecast more quickly and accurately the financial impact of any company decision to add or subtract flights from its schedule. It also needed to be able to forecast the impact on American of other carriers’ decisions to add or change routes that are in competition with its own service.
In this new competitive environment, several large airlines did fail. Pan American Airways, the great pioneer in overseas passenger flights, was the most conspicuous example of an airline that finally had to shut down. This development made it clear to the American Airlines management team, including CEO and chair Bob Crandall, that accurate and speedy forecasting was urgent if the company wished to remain successful. Unfortunately, forecasting systems in those days consisted only of PC spreadsheets and calculators. Using low-level technology, American analysts had to work for days merely to estimate the effects of a change in only one single flight. Forecasting the financial effects of a large schedule change was well beyond the abilities of American’s planning department. American soon moved to develop a powerful mainframe-based forecasting system that is named Integrated Forecasting System (IFS).The 1989 collapse of another major competitor, Eastern Airlines, gave American the opportunity to test its new system. During those years airlines were very busy building hub systems, and to ward the end of the decade, American had decided it needed to build a new hub somewhere in the southeastern United States. Atlanta, Georgia, emerged as the leading candidate for the location before the Eastern bankruptcy became public. However, the Eastern collapse opened up another possibility because Eastern Airlines had dominated the profitable New York-Florida routes. Now American had to consider Miami, Florida, as another possible hub location. Management at American quickly turned to its newly completed IFS for supporting data to help the management team make a decision as to which would be the besr site for their new hub. The problem for the IFS was that this decision was infinitely more complex than that of making a change in the schedule of a single flight. A new hub meant a dramatic change in many flights and routes, and the new IFS would have to forecast the results of all these massive changes. Moreover, it would have to it for two separate locations. The planning department put IFS to its first big test, and it passed with flying colors.
To complete the whole task IFS only had to work a total of eight hours। The system output indicated that Miami would be more advantageous financially, and Crandall quickly presented that plan to his board of directors who then approved it. According to Brad Jensen, vice president of application development for Sabre Technology Solutions, the Miami hub has generated hundreds of millions of dollars in profit in the years since that decision
The new IFS system did have a major problem-it had to handle am immense amount of data. The system draws on 60 source data files, supplying both internal and public data. The files provide such varied information as passenger demand, fares, aircraft seating capacities, unit costs per flying hour, and even competitor flight schedules. To try to understand the quantity of data, consider that American has about 4,000 flights daily, and that the company must track at one time approximately 400,000 flight combinations (that is, pairs of cities that are or might be linked by flights). Moreover, the success of IFS on the hub questions generated another problem for American. The demand for the use of the system rose rapidly. However, users were unwilling to wait eight hours to get a response to their questions. In order to speedily handle such an enormous amount of data while responding to the greatly increased number of requests, The Sabre team decided the system would have to be moved to a parallel-processing architecture. A project to accomplish this goal was estimated to cost an immense $16 million. Fortunately, Crandall supported the project because the SABRE reservation system had been so successful. In fact, according to reports, when Crandall first was told of the proposal, including the cost, he jumped out of his chair yelling, ‘’I want the system.’’ Sabre management also was strongly in support of the project. Thomas M. Cook, president of Sabre Technology Solutions, remembers that, ‘’There was a realization among the senior management at both companies that in order to have significant breakthroughs, you have to take significant risk.’’
The development team made two critical decisions in 1991. The first was to convert to a client/server platform that would not only provide a parallel-processing environment but would place control of the upgraded system on the desktops of the users. Thus, the users would be able to initiate and manage their own forecasts right from their own PCs. The second decision was to adopt a new and, at that time, relatively unproven technology, object oriented programming. On reflection a third decision has been seen as critical to the success of the project. The development team decided it must work closely with the end users, American’s capacity planning department, and so the team was housed on the same floor as the planners. The result was a very close working relationship between the two sides and a successful project. Adding to the project’s success was the commitment by senior American executives to attend weekly project meetings, thereby lending not only their strong and visible support to the project but their design and usage insights as well.
The project was completed in 1996, and American found itself with a much faster and more powerful IFS. Sabre continued to improve the system so that the queries it can handle today are much more complex even than were those in 1996. It now forecasts within 3 minutes the company’s market share for its 4 million services worldwide as well as 400,000 city pairs. With the new client/server system, planners can actually get answers to hundreds of questions about a potential single flight change, whereas, previously, the mainframe IFS could only handle two or three questions. American planners are now able to do much more planning, do it more effectively, and yet do it with a reduced staff. American’s planners use IFS constantly to find ways to cut costs and increase revenues. The planners analyze the financial impact of adding or deleting flights, adding service to a new city or dropping an existing service, or changing the size of an aircraft for a specific flight or route. They even use it to project the financial impact of schedule changes made by American’s competitors. One interesting example of a result of IFS projections was American’s decision to replace some jet planes with propeller planes. Despite the fear of many that the change would result in a loss of passengers, the system showed that American could redeploy the jets by adding flights to longer, more profitable routes. American’s cost-per-mile declined while profits rose. All-in all American planners and management have gained a much deeper and more complete understanding of its customers, its competition, and the market place. Moreover, management claims American’s profit have gone up about $500 million as a result of the first 2 years of using IFS.
Although American Airlines has benefited greatly from AMR’s ownership of Sabre, that relationship is about to change. Over the years Sabre has developed many systems for other companies, including selling versions of IFS to the London subway system and to the French railways. However, Sabre’s close ties to American Airlines have been seen by other airlines as a problem. According to James Poage, senior vice president of worldwide marketing for Sabre, the problem ‘’we faced in the past is [other airlines would] say ‘Because of your relationship with American Airlines, is our information secure?’ ‘’ Because of this fear Sabre garnered no significant airline outsourcing contracts during 1988 and 1999. As a result, and with the goal of increasing the potential for Sabre, AMR announced on December14, 1999, that it is spinning Sabre off, making it a fully independent technology company. Sabre president and CEO Williams J. Hannigan explained the expected result, saying ‘’with our independence, it’s more clear that success for Sabre won’t mean enrichment for American. That’s real important to those carriers. Soon AMR will be able to sell its systems to any company and will be free to develop for any airline.’’
Sources: Scott McCartney, ‘’Airlines Find a Bag of High Tech Tricks to keep Income Aloft,’’ The wall Street Journal, January 20, 2000; Stacy Collett, ‘’Update: Sabre to Sell Software Once Controlled by American Airlines, ‘’ Computerworld, December 14, 1999;Polly Schneider, ‘’clear Skies Forecast,’’ C10 Magazine, February 1, 1999; Lauren Gibbons Paul, ‘’From Harrowing to Heroism,’’ C10 Magazine, September 1, 1998; and Linda Rosencrance,’’ AMR to spin off Sabre, ‘’ Computer world, December 12, 1999.
CASE STUDY QUESTIONS
- Evaluate American Airlines using the value chain and competitive forces models.
- How is forecasting related to American Airlines’ business strategy? What kind of DSS is IFS?
- How did IFS help American Airlines pursue its business strategy? What kind of DSS is IFS?
- What management, technology and organization problems did the IFS project present to American Airlines? To Sabre?
- What were the advantages and disadvantages for American Airlines of having its Information Technology development being done by Sabre, a sister company, rather than by American Airlines itself? How do you think the spin off Sabre will affect the competitive position of American Airlines in the short run? In the long run?
