Looking Out for No. 1

By Peter Horner

Hey, Tom Cook. You built the biggest, most influential operations research outfit in the world at Sabre. You oversaw the development of the quintessential OR application known as yield/revenue management that revolutionized the airline industry. You then joined McKinsey and Company as a senior counselor, started your own company called Replane Inc. and successfully ran for the presidency of INFORMS. Along the way, you helped your old team at American Airlines and Sabrethe Super Bowl of Operations Research — the Edelman Competition — not once, but twice. What are you going to do now?

"I'm going to CALEB!"

OK, those aren't Cook's exact words, but he is going to CALEB Technologies Corp., an Austin, Texas-based software solutions company located just a short commuter flight from Cook's old Sabre stomping grounds at DFW. In fact, Cook's already on the ground in Austin. When CALEB founder Gang Yu announced his retirement in January, he handpicked Cook — whom Yu describes as a "legendary business leader, especially for operations research and information technology professionals" — to "take CALEB to the next level" as chairman and CEO.

Did we say next level? How about the top floor? When Cook gets a hold of an OR/IT group, he aims high. Given his history with Sabre and the airline industry, you might say the sky is literally the limit.

"Our objective is to become the world's leading provider of decision-support solutions," Cook proclaims in an interview as he is still settling into the CEO chair at CALEB. "By decision-support solutions, I mean melding OR with IT to solve a problem for an organization in the private or public sector. That's the vision."

CALEB (Computer Applications in Logistics Engineering and Business) provides disruption recovery and resource planning systems for complex, mission-critical applications. At the moment, CALEB is focused on the airline industry — Continental, Northwest, Southwest and AirTran are major clients — but Cook and Yu both see the company expanding into a wide range of industries in the years ahead.

Won't that shared vision put CALEB in direct competition with Sabre, the mother of all OR groups; the group Cook presided over for nearly two decades; the group that produced an estimated $1 billion in annual incremental revenue for parent company AMR and its subsidiary, American Airlines, by the time Cook left in 1999; the company that now markets its portfolio of software solutions to a host of other industries including hotels and cruise lines; the company that sets the standard by which all other OR groups are measured?

The thought draws a chuckle from Cook, followed by the kind of response a football coach might give on the eve of the big game against a friendly in-state rival.

"We're already in competition with Sabre," Cook says. "And that's just fine. We've already gone head-to-head with them on potential business. I have a lot of friends over there, people I respect and admire, so it's a friendly competition, and one that I welcome."

Cook says he was drawn to CALEB for many reasons, including his longtime friendship with Yu, the company's "strong team," its "in-depth understand of airline operations" and especially its "world-class" software.

"CALEB is exactly the kind of company I want to run and grow," Cook says. "It's an operations research-based company where the competitive advantage of the product they sell is based on value that we all know OR brings to these kind of problems. They have a good client base. And they have state-of-the-art solutions. They solve a problem that no one else in the world has solved. They are way ahead of the competition in their particular space."

That space — disruption management — is getting a little crowded. ORMS Today recently documented the work of a group of researchers at the Department of Informatics and Mathematical Modelling at the Technical University of Denmark who are pursuing disruption management solutions in conjunction with British Airways and other partners [1, Clausen et al.]. Of course, Sabre, the 500-pound gorilla in decision-support solutions, always takes up a lot of space.

Just to make the budding rivalry more interesting, Cook has already hired some former Sabre staffers. It should be noted that in 1982 when Cook took the reins of the OR group (that would become a part of Sabre) at American Airlines, he inherited a staff of about a dozen OR/IT professionals. By the time Cook left Sabre in 1999, the company had grown, through new hires and mergers, to more than 5,000 professionals and was widely considered THE "world's leading provider of decision-support solutions" [2, Horner].

Will History Repeat Itself?

As of mid-January, CALEB had a total of 72 employees, including a software development operation in Shanghai. Can Cook repeat his OR magic at CALEB? Only time will tell, but there are obvious parallels between the two stories, starting with the man in charge.

"How we get there and how fast we diversify out of the airline business, we'll see as we go," Cook says of his vision quest. "I suspect in the near term we'll stay in the transportation industry. Once we have more traction in transportation, we'll start branching out.

"The same thing happened at Sabre when we built AADT," he continues. (American Airlines Decision Technologies, or AADT, eventually merged and morphed into what is now known as Sabre). "We started out in the airline business and lots of opportunities came to us, and we pursued other opportunities outside that particular space. Many industries need these kind of solutions, not just the airline industry."

The transportation sector in general, and the airline industry in particular, has always been fertile ground for operations research because airlines are confronted with huge, complex problems that demand decision-technology solutions. Flight and crew scheduling are classic examples, along with the pricing dilemma brought about by deregulation in the late 1970s: How could major carriers compete for passengers with low-cost airlines that can operate the same route at a fraction of the cost?

Looking back, the answer might seem obvious, but it revolutionized the industry: Offer a portion of your seat inventory at matching prices and save the rest for late-arriving, high-yielding demand, i.e., business travelers. Dynamic pricing was the answer, but some tough questions remained: How many seats do you sell at a discount and how many do you hold in reserve? When do you change the price of each "bucket" of inventory and by how much? Those questions gave rise to the concept of revenue/yield management, perhaps the most significant OR application of the last 20 years.

Even when revenue management offers a relativelypercentage gain in incremental revenue (5 percent to 7 percent is widely considered the industry standard), the payoff is huge when you're talking about an airline with annual revenues in the neighborhood of $15 billion. It's called leverage, lots of leverage.

Today, thanks to the work pioneered by Cook's troop at American Airlines and Sabre, virtually every airline in the world employs some sort of revenue management system. The concept is also hard at work throughout the cruise, hotel and car rental industries, and wherever else you find perishable, limited capacity. [3, Horner; for more on revenue management, see "Why CEOs Should Care About Revenue Management" of this issue].

Ironically, Cook and company couldn't come up with a feasible solution to the disruption management problem during his Sabre stay.

"We tried to solve this particular problem a number of times but we were never successful," Cook says. "So I have a lot of respect for the team here. The set of solutions CALEB came up with is unique. They solved one of the problems that got away from me in a previous life."

Disruption Management & Recovery

Like the crew-scheduling problem, the disruption management/recovery problem involves many individual considerations that are coupled in complex ways. The major players include crew, aircraft, passengers and maintenance. Decisions made about one set of players impact others.

Following a significant disruption — typically a weather-related closure of a hub airport — the affected airline wants to get back on schedule as quickly as possible at the least possible cost. The airline also wants to get its paying customers to their destinations without needlessly jeopardizing passenger goodwill. Sometimes the objectives are in conflict. Canceling a flight may be the most cost-effective solution, but what does that cost in terms of passenger goodwill?

The problem is made more complex by factors beyond the airline's control such as the weather, gate availability and runway capacity. "Airline operations require many resources, and these resources are very often tightly coupled," Yu says. "Without a crew, without gates, without runways, without aircraft, without maintenance, an airline cannot carry out a flight. During the schedule planning stages, all of these resources are allocated very tightly, often independently, without considering the real-time operations. That's why a disruption will have a snowball effect throughout the network, and the impact is usually pretty big."

The hub-and-spoke scheduling favored by most airlines virtually assures that a disruption of the hub will quickly create waves of problems throughout the system. "You're talking about a very tightly wound network," Cook adds. "It's tightly wound because the industry is so competitive. Of course, you could solve a lot of these problems by having every flight go point to point, and doing away with connecting hubs. You could mitigate the problem by putting a lot of slack in the schedule. But as soon as you do those things, you start hurting yourself so bad from a profitability standpoint that it doesn't work. So all airlines are pressured into designing very tight schedules and that makes them inherently unreliable."

Last but certainly not least, the recovery problem is not only huge in terms of variables, it's one that demands near-instant answers in order to be effective. Meanwhile, the clock is ticking and the money meter is running.

A hub disruption typically costs an airline between $2 million and $3 million a day. If Atlanta's Hartsfield International Airport shuts down — as it did for several hours one day last fall when a passenger breached security — Delta and AirTran immediately feel the financial pinch.

The hard-dollar costs of disruption are relatively easy to calculate. The softer costs, such as lost passenger goodwill, are almost impossible to measure. Both are important. Along with tallying up lost revenues and the incremental cost of the disruption from an operations standpoint, an airline has to be concerned about how its passengers believe it dealt with the situation.

Most disruptions are weather-related; snowstorms shut down airports in the Northeast virtually every year. Continental Airlines, for example, experienced four major disruptions last year in addition to the 10 to 20 minor disruptions that impact operations on a daily basis.

Three of Continental's major disruptions last year involved bad weather, such as snowstorms that closed down its hub at Newark (N.J.) and flooding in Houston. The fourth, of course, occurred on Sept. 11. The terrorist attacks upped the ante substantially and thrust the concept of disruption management and recovery into the national spotlight. Shortly after hijacked jetliners crashed into the World Trade Center and the Pentagon, President Bush ordered the national airspace shut down.

When the ban was lifted three days later, planes that had been diverted were not where they were supposed to be. Neither were the crews. The airlines were faced with a situation they had never seen before. How could they restore their flight schedule and minimize the impact — to their passengers, to their crewmembers and to their bottom line? Continental Airlines turned to CALEB's OpsSolver and CrewSolver software for answers and, according to company officials, made a quick, efficient recovery from a worst-case scenario (see sidebar article "How Continental Landed On Its Feet").

Inside the Black Box

Instead of formulating the problem as a giant network problem, CALEB's software employs a combination of OR techniques — fuzzy logic, intelligent search methods, classification of constraints and prioritization of objectives. Decomposition methods further reduce the problem — which can involve hundreds of thousands of discrete variables — to a workable size, allowing it to be solved in a matter of seconds or minutes.

All of the considerations of the recovery problem — lost revenue, passenger goodwill, cost of operations, crew, maintenance, etc. — are mathematically weighted in different combinations. The weights can be adjusted depending on the airline's particular objectives, management practices and business rules. The software then presents least-cost options to fit the set of objectives.

Doesn't the passenger always get the short end of the stick? Not necessarily. Cook points out that what is best for the airline and what is best for the passenger in disruption situations are not mutually exclusive. In fact, they are often linked.

"The airline is trying to minimize the loss of customer goodwill AND minimize the loss of revenue," Cook says. "When they have to cancel a flight to balance their schedule and they have the tools we're talking about, they're going to look for their least-cost flight. Typically, that's the one that has the fewest number of passengers or the flight that causes the fewest number of interruptions in passenger itineraries. The two are not 100-percent correlated, but they are highly correlated."

Airlines may have a different set of objectives on different days of the year. For example, an airline might want to maximize customer goodwill and do everything within its power to get passengers home the day before Thanksgiving. Toward the end of the fiscal quarter, however, when an airline is trying to make its financial target, it may want to minimize the cost of recovery. CALEB's software is designed to conform to an airlines' objectives — short-term, long-term, even on a day-by-day basis.

Cook, of course, is no stranger to the kind of decision-making that goes on inside an airline boardroom. As the head of Sabre Decision Technologies, Cook sat in on many such meetings with legendary AMR boss Robert Crandall throughout the 1980s and 1990s as Crandall guided American to the top of the industry.

"You don't work with someone as close as I did with Bob and not learn a lot," Cook says when asked to compare his management style to his former boss at AMR. "Bob is a very unique and incredibly talented chief executive. I admire him tremendously. As for my own management style, you would have to ask the people who work with and for me. I consider myself pretty much a hands-on guy and I share that with Bob. I look for the right people and give them the authority to get things done. I'm results-oriented."

As the CEO of CALEB, Cook will be looking to grow the company with "smart, driven, passionate people," the kind of people he surrounded himself with at Sabre.

"Obviously, I have a real bias toward OR people," he says. "On the technical side of the house, we will try to hire the very best OR and IT people we can find. Any successful business has to have a lot of skill sets other than technical. Many of those skills are already here, and we will recruit more. I met many very smart people with good business skills at McKinsey, and they could make a positive contribution to a company like CALEB."

When Gang Yu founded CALEB seven years ago, he envisioned an OR-based company that would grow to become the world's leading provider of optimized business solutions to a wide range of clients. "The algorithms and models we've developed have applications in many different fields, including supply chain and telecommunications," Yu says. "We have a foundation for general, real-time operations control and disruption management."

Clearly, Yu and Cook share the same vision. Now it's up to Cook to quarterback the team, perhaps throw a few touchdowns andanother Super Bowl of Operations Research.

Is he up for the challenge?

"I'm looking forward to it," Cook says. "It's a big challenge. I know, having gone through this before at Sabre. I'm not trying to prove anything. I'm just trying to have a good time."

And how will he measure success?

"We will measure our success by whether or not we realize the vision. Remember, the vision is to become world-class in a very broad sense. The goal is to become the world leader."

If Cook pulls that off, he won't be just the MVP of the Super Bowl of Operations Research. He'll be headed for the OR Hall of Fame.


- Clausen, Jens, Jesper Hansen, Jesper Larsen and Allan Larsen, "Disruption Management," 2001, OR/MS Today, October, p. 40-43.
- Horner, Peter, 2000, "The Sabre Story: the Making of OR Magic at AMR," 2000, OR/MS Today, June, p. 46-47.
- Horner, Peter, 2000, "Mother, Father of Invention Produce Golden Child: Revenue Management," 2000, OR/MS Today, June, p. 47.
Be sure to read the sidebar article "How Continental Landed On Its Feet."

Peter Horner is the editor of OR/MS Today.