Delta Aims for Infrastructure Overhaul

The sepia-toned photos lining the walls of Delta Air Lines’ Technology office inadvertently tell the story of an airline mired in tradition: reservation agents from the 1950s shuffling through 3-by-5 cards, crew-cutted control tower personnel sorting through sheafs of paper. While reservation agents are no longer working off index cards—and haven’t been for some time—Delta Air Lines has been known throughout the aviation industry as a technology laggard. As little as five years ago, much of the company’s operations remained paper-based, and Delta was one of the last major carriers to invest in the Internet.

But in 1998, the 78-year-old airline embarked on an enterprisewide shift. In an attempt to catch up with industry peers like the largest U.S. carrier, American Airlines—an IT pioneer that created Sabre before any airline had a computerized reservations system—and number-two United—a fast technology follower—number-three Delta began a multiyear, billion-plus overhaul of its IT systems. Thus far, the Atlanta-based airline has spent almost $1.5 billion on technology in the past several years, with more than $50 million of that total going to what it calls the Delta Nervous System (DNS), an entirely new infrastructure on which it has been building applications that automate everything from tasks done at the gate and boarding area to baggage-handling, and inventory and revenue management.

Creating a new technology backbone and building all new applications on top of it—while continuing to run an airline—is no easy task in any environment. But the project has also met with a harsh dose of reality along the way. Though work on DNS began in a time of prosperity and stability, $13.9 billion Delta Air Lines has since hired a new CEO, slashed more than 20,000 jobs and gone through three CIOs, the economy has tanked, and America suffered the worst attack in U.S. history leaving the airline industry reeling with bankruptcies, belt-tightening measures and beleaguered major carriers partnering up in an effort to save themselves. “It has actually been like trying to change cars while driving down the highway at 100 miles per hour,” says Henry Harteveldt, a San Francisco-based travel and airline analyst for Forrester Research. “Oh, and by the way, one of the cars just blew a gasket.”

Many companies would pull over and turn off the car—or even put it on blocks in their garage. And the longer the economy suffers, the harder it will be for companies to stay their course. At a time when some airlines have shelved expensive IT efforts, Delta has continued to invest in more infrastructure and applications, albeit on a delayed schedule.

Delta, which made $990 million in 1998 when the work on DNS began, lost $1.2 billion in 2001 and $1.3 billion in 2002. Last year, it spent $200 million on new IT development (down from $218 million in 2001 and $371 million in 2000) even though total capital spending for Delta decreased by a billion dollars last year. “Delta continues to see this as a key investment,” says Curtis Robb, Delta’s current CIO. “And Delta Technology will play an integral part in returning Delta Air Lines to profitability.” Analysts add that continued investment has the potential to eventually make Delta a leader not only in technology but in the airline industry overall.

But if and when that will happen remains to be seen.

No Choice but to Start Over

Through the mid-’90s, much of Delta’s operations ran on spreadsheets and checklists. Pneumatic tubes were used to shuttle information throughout airports. What systems were in use were scattered and disconnected.

In 1996, things started to go from bad to worse. An outsourcing plan flopped, and so in 1997, the airline hired Charlie Feld, CIO-for-hire and turnaround expert, to heal the technology group.

“They had projects in disarray, people in disarray,” remembers Feld, now CEO and founder of Irving, Texas-based Feld Group. Departments at Delta were buying their own technology and hiring their own programmers without any coordination to avoid duplicating efforts. At the same time, the airline had to figure out a way to cope with the pending Y2K problem.

Feld immediately created the wholly owned Delta Technology subsidiary to provide the airline with IT services. And his team decided that the best way to deal with the disparate Delta systems and the year 2000 issue was to rip out the existing, antiquated infrastructure and start over. And while they were at it, they figured, they could start building new applications on which to run the airline. It was a bold move in an industry in which many companies have chosen to live with legacy systems rather than face the costly, complex task of coming up with an enterprisewide technology solution for replacing them, according to John Linehan, an analyst at Baltimore-based T. Rowe Price.

But inside Delta, it was viewed as the only move. “We had seven to 10 years of technology that was just getting older and older, and the risk of it being unreliable was getting bigger,” says Dean Compton, vice president of common systems for Delta Technology, who’s been at the company since late 1996 and helped get DNS off the ground. “We figured we might as well replace it with something we could build on for the future.”

It was sound reasoning, and in the strong economic times of the late 1990s when Delta was selling a lot of expensive tickets, it didn’t take much work to convince leadership that it was the way to go. New CEO Leo Mullin even made it one of his main initiatives as the new leader of the company.

Mission Delta

Every 40.7 seconds, a Delta jet takes to the sky. Getting that airplane off the ground takes information from flight schedules to gate information to baggage-handling to tower operations to customer service. Delta’s 2,123 flights each day also require 7.3 million gallons of fuel, 109,000 meals and snacks, 151,000 bottles of water, 87,000 cans of soda, and the list goes on.

“It’s basically a big, real-time manufacturing operation,” says Robb. “And the one thing that’s unique about it is that the customer is actually right in the middle of it, so when it doesn’t work well, it’s very visible.” Another distinction of Delta’s manufacturing operation is that accurate advanced planning is nearly impossible. Everything from weather to shifting economic realities to maintenance issues put wrinkles in plans every day.

So to get much-needed real-time information, Delta chose infrastructure software from Palo Alto, Calif.-based Tibco Software, known for providing systems for Nasdaq and the New York Stock Exchange.

Unlike the TPF (transaction processing facility) systems Delta’s airports formerly ran on, which required users to type in a cryptic string of letters and numbers to retrieve information every time they needed it, this new system would operate in front of TPF using a “publish and subscribe” architecture. It would recognize when a certain event occurred—a plane arrived at a gate, a passenger checked in—and would automatically push that current information to the people who needed it.

Construction began in fall 1997. Feld chose to begin by building the gate and boarding application and the infrastructure necessary to support it. That application would touch the most areas within Delta, be most visible to its 104 million passengers, and force Delta Technology to collapse most of its 35 existing customers databases and 40 flight databases (along with a slew of others from crew management to catering). Once that was done, they could create 10 new common data stores, get the necessary middleware in place and build a good chunk of the DNS infrastructure that other applications could share. Fourteen months later, the gate and boarding application rollout began.

The critical first step was a success. Once gate agents got used to the new system, which gave them current information on everything from who had checked in and the location of the connecting passengers to a layout of seats and assignments and standby status, it saved an average of eight to 10 minutes per flight and reduced training time. “I remember when we put in the system in Jacksonville. I saw an overbooked widebody 767—here there’s normally a lot of confusion around the gate—boarded by two agents using the technology on time and ahead of schedule,” Compton recalls. “I’d seen a similar situation in Salt Lake City where we hadn’t put the technology in yet, and they had to use nine agents to board the plane, and it still left late.”

In 1999, Delta Technology continued to replace the hardware and wiring in its airports, and began to develop and roll out more applications to plug into DNS, from gate information display systems for passengers to a baggage-transfer application used by luggage handlers via a rugged laptop. Feld thought things were moving along well enough to turn things over to his successor, Robert DeRodes, who came from Citibank, as well as a CTO, Robb, an IBM veteran.

Getting Nervous

After becoming CIO in December of 1999, DeRodes continued the DNS work. Among the work done in 2000 and 2001: Check-in kiosks were built, Delta.com was redesigned, flight-searching software was acquired, flight-information display systems were replaced, and virtual check-in was introduced. All the new applications were wired into the new, ever-expanding infrastructure.

Staying the Course

Technology spending is often an easy mark for budget cutters when profits tank. But Delta continues to invest in the Delta Nervous System, a sign of its deep commitment to the technology upgrade. Learn more.

But by early 2001, the airline industry had begun to feel the effects of the softened economy. The boom times were over for now, and blindly investing in IT wasn’t going to fly. That became clear that spring at a meeting of Delta Technology’s board of directors, made up of Delta’s president, new CFO Michele Burns, the chief marketing officer and the executive vice president of HR. More than two years into the DNS work, the executive team was becoming impatient. “I remember the exact day—April 17,” says Robb, then CTO. “They said, What in the world is this project called Delta Nervous System, and why in the world do you want to spend millions more on it?”

DeRodes and Robb presented a three-hour overview of the project—what was done, what they wanted to do and why. Lucky for them, Vicki Escarra, the chief marketing officer, was in the room. She had been in charge of airport customer service when they installed the first DNS application, the gate and boarding system. “We had her saying, Man, we couldn’t have done what we’re doing today if it wasn’t for this technology,” Robb recalls. The combination of the overview and the satisfied customer from marketing seemed to do the trick. Each project that either lowered operating costs or increased revenue was given the go-ahead. But Burns booked the potential savings and revenue into the budget to show how serious they were going to have to be about getting the returns they were promising, and all future projects had to go through a business case justification. Says Robb: “At the end of the meeting, they looked at us and simply said, Speed it up.”

But just as Delta Technology was attempting to pick up the pace, the airline industry skidded to a halt on Sept. 11. In October 2001, Delta Technology’s board reconvened to look at DNS projects with an even more critical, cash-conscious eye. The end result was putting off some projects that had a longer payout (new HR systems and rolling out the baggage-handling system in additional airports like Cincinnati and Salt Lake City) and speeding up others that would inject immediate cash into the company or cut costs right away (drastically increasing the number of self-service kiosks and replacing call center technology). Project priorities shifted, but the overall commitment to new technology remained.

“[Delta’s leaders] view technology as one of the mechanisms for speeding the recovery, and they decided to stick with those investments,” explains Robb, who took over as CIO when DeRodes left to become CIO of Home Depot in February 2002.

Altering Course

Currently there are more than 20 applications running on the new infrastructure, and Delta processes more than 5 million daily and 134 million monthly business transactions using DNS. Last year, Delta made a commitment to tie in SAP for inventory management, replace technology and introduce DNS-powered software at its 12 major call centers, and develop applications for revenue management that will require significant additions to the existing DNS backbone. But the “flight plan” for DNS and its associated applications has definitely been altered. “Today, we’re working only on projects that pay off in 12 months and have ongoing impact for at least three years—either building revenue or lowering operating costs,” says Robb.

And certain aspects that may affect the ultimate success of Delta’s new commitment to IT remain out of Delta Technology’s control. “A big part of how fast we can do this or how successful we are is the airline industry itself. There are always outside factors—like the economy—that have an impact on what we’re doing,” says Compton. To deal with the decreased capital, Delta Technology is scaling back on rollouts. New and existing applications may be introduced in just one or two hubs instead of all 40 airports, for example, with the intent that once the financial situation turns around they’ll be ready to speed up the rollout.

The good news for Delta Technology is that Delta Air Lines’ balance sheet looks better than most major carriers. While its operating performance is similar to its big hub-and-spoke brethren and its losses place it at the middle of the pack, it has the most cash of its peers ($1.7 billion in the third quarter of 2002), giving it more staying power and the ability to continue these kinds of investments, according to Jamie Baker, an airline analyst at New York City-based investment bank J.P. Morgan Chase.

Delta says it is likely to continue to lose money this year, putting into question how the billion-dollar DNS buildout has really boosted the bottom line. “It’s like they’re in a boat that’s taking on a lot of water,” T. Rowe Price’s Linehan explains. “These new systems allow them to bail faster, but the water is still coming in faster than they can bail. Clearly there are lots of benefits to these systems, but it’s more difficult to see them right now.”

Nonetheless, the current plan is to keep working on DNS-related projects for the foreseeable future, perhaps creating a new kind of tradition for Delta and Delta Technology. “This is not one of those quick fixes or a quick journey,” says Robb, noting that some of the most complex areas to connect to DNS, like revenue and inventory management, still lie ahead. “When it comes down to it, I see this work going on for another five years. And that’s only because I don’t know how to make a plan any further out than five years. We’ll be done when we run out of ideas.” But then again, even the best plans have been known to change.