Airline IT cutovers are hugely complex, and often end in disaster. Why? And what can airlines do to avoid the pitfalls that plagued their predecessors? Report by Gillian Jenner in London
Notoriously risky, tying up vast amounts of resource, with the cost to balance sheet and brand potentially going into freefall if it goes wrong - you would be hard put to overestimate the stress passenger service system cutovers place on airlines.
These events tend to happen once in a generation - anything from 10 to 30 years plus, depending on the airline. But sooner or later, changing strategy, business model, expansion into international markets, mergers or takeovers, joining a global alliance or obsolescence in the current system will impel airlines along this daunting path.
This is not just an IT or technology issue. It is also a people issue. Yes, successful cutovers are founded on meticulous attention to detail, lots of checking and double-checking that the new system will work with all the other applications and that you can transition without compromising data. But they also demand wholehearted engagement at every level of an airline - from the check-in agent to the chief executive officer. That is a really challenging combination.
Take too long or spend too much on your cutover, and greater business pressures will lead to the plug being pulled at significant cost. WestJet wrote down C$31.9 million ($31 million) in 2007, when it scrapped plans to develop and implement the tailor-made aiRES reservations solution. And last year, Air Canada suspended work on the next-generation Polaris res system it was developing with ITA Software, which had chalked up costs of some C$67 million, to save C$40 million in migration costs to take it live.
Why are these cutovers so hard to execute? Other industries seem to pull off massive system migrations without the kind of problems that plague our industry. Airlines like to think they are unique - and, in this instance, they may be right. Their high-volume, high-value product is perishable, and is on sale 24/7 in an extremely dynamic marketplace.
Many banks have old host business transaction systems with the same level of complexity. However, Lufthansa Systems executive board member Gunter Küchler points out: "They don't do transactions at night. As opposed to other industries, airlines are doing transactions every minute of the year, everything has to run immediately for the whole global market."
The other issue is that airlines are not just swapping one system for another. They are changing processes. "Airlines are changing their business. In other industries, they are doing more or less the same for many years. New systems are not implemented because of business or process change, " says Prologis executive director Gerd Pontius. "Whereas airlines have to deal with business change and system change."
Now, add in the complexity of the reservation system having connections into hundreds of other systems and applications. A host passenger service system (PSS) can be connected into flight inventory and schedules, a pricing component and a revenue management tool to help maximise yields. Then there are the connections into all the sales channels - web, call centre, travel agents and GDS - plus, if you run a loyalty programme, there are connections into the customer relationship management system. The host reservation system will also need to talk to the departure control module, and that may be a third party module if a ground handling company is involved. It also has to have capacity to tie flown revenue back into the revenue accounting system.
"The fundamental reason it's not easy is [that] it is the business equivalent of taking out the brain of a human being, while they are laying on the table and you're having a conversation with them," says Chris Vukelich, ex-British Airways general manager of global distribution and chief operating officer of Travelport, who is now a senior vice-president with payment solution firm eNett.
Every year, the PSS gets older. Every year, the system is tweaked and additional bits of functionality are added, and fewer people in the organisation really understand in detail what these systems are doing. Not only do airlines underestimate the complexity of the migration, they tend to overestimate their own capability, in terms of resources, time and knowledge, cautions Prologis's Pontius.
"Very few people have an understanding of the crossover process of the whole organisation. For instance, how the call centre crossover will affect revenue accounting," he says. "There are very few champions. They are always short of time, and are required by most projects in airlines."
That lack of knowledge can also extend to airline people not fully understanding what migration means, nor being completely clear about what they need to do. This is where the host providers, who are handling cutovers on a regular basis, need to make a judgement call on whether the airline is mature enough, and has the right people to support it.
Ursula Silling, who is now a consultant at XXL Solutions, was formerly vice-president, commercial with Brussels Airlines and chief commercial officer at Aerosvit. She says some providers do not consider that more resources may be needed to support airlines in regions where systems know-how and English language skills may be lacking. The systems are all in English, and many people working on them may not be able to fully follow what needs to be done. Cultural barriers prevent people from admitting this, Silling explains: "This problem must not be underestimated."
Citing examples from Italy, Spain and Russia, she adds: "Quite often, a lot of specialists don't speak English." Silling says this leads to airlines excluding IT companies that only do business in English, because they cannot complete the contract process. They then end up with a local product and provider. The result? "The risk of failing is even higher."
Providers are in a good position to steer airlines away from simply replicating their legacy system in a new environment. Amadeus vice-president for industry affairs, Hans Jorgensen, says: "When we have done gap development for an airline to replicate what they had in the past, very often, 40% of this is not used after the migration." This is because staff find the new system better. "If you accept it's going to be different, and put all the effort into training, they will jump into it."
Treating the PSS cutover as a business project should be the overriding principle. "This is not an IT project. This is a fundamental change in the way the airline will deliver service to its customers and a proper level of governance within the airline," says SITA vice-president for airline and passenger solutions, Brian Cook.
Jaime Pocasangre has a PSS migration at TACA under his belt and, as chief information officer, is now readying Aeromexico for the switch to Sabre. He agrees this is a business process transition: "It is not just training in functionality, it is also training in the business process. It is not the system that will make the difference. It's the combination of that business process with the system. That is a fundamental principle."
The TACA migration also taught him that it is the people involved that count. "The number one thing I learned for any migration is that people make the difference, so the ability to have top resources in key positions of the project, that's the difference for successful implementation," he explains.
That involvement of senior resource should go all the way up to the very top of the airline. "You need to have a steering committee that represents most, if not all, of the stakeholders. And that steering committee needs to be active - this cannot be a second day job," adds SITA's Cook.
Ownership from the senior leadership is essential for a successful cutover, agrees eNett's Vukelich. Ideally, the migration sponsor should be the airline chief executive, "who needs to pay attention and to pay more than lip service to how the project is going", and empower the project manager to get things done. "Without senior sponsorship, the thing could be a train wreck," he says.
Active involvement from people around the business is also vital. Familiarity with the old systems and its short cuts means some departments may either assume change will be easy, and not pay attention to their part. Or they may say: "Bring in the new system - but you're not changing my process."
American Airlines signed up earlier this year to work with HP on the next-generation PSS to replace the Sabre system that it pioneered. Generations of employees have grown up with the 45-year-old system, so how do you start getting buy-in? "We're years into it," explains American Airlines chief information officer Monte Ford.
"We spent several years on the business complexities of such a change in the environment, and spent a lot of time talking to people in the business units - road shows, discussions, open meetings, blogs, focus groups. We spent a lot of time understanding what people need, and what they think they will need for the future." This has included thinking not just about what the airline needs from its PSS technology, but also the technologies its customers use, and how technology will drive the way business is done.
Now Ford and his co-leader on the project's business side have a different kind of predicament - holding back the excitement. "We have got tremendous buy-in and anticipation for what we are going to do, so now we have the opposite problem. We're telling people to hold on - we're not going to do it overnight."
This engagement is also key to helping the carrier manage its project resources and outcomes during the three to four years it will take to build the new system. "It is really important to us that our employees and the companies we work with understand we cannot afford to have large spikes in spending or massive disruption in our operation for weeks on end," he says.
The time an airline takes over a migration project is critical. And, with so much potential risk, it is inevitable that a significant chunk of time is invested in the migration decision itself. LAN first started considering its IT migration back in 2006, and it was not until 2009 that it finally made the decision to transition from its Amadeus-Resiber system to Sabre after evaluating three alternatives.
"Host systems represent one of the biggest expenses of an airline," says Sergio Mendoza, LAN's vice-president of distribution and revenue management. "You spend quite a lot of money and teamwork [on this], so you are likely to look at it frequently. You don't change the host system frequently, but you're always looking at it."
Work started in April on scoping out the project and its milestones, and Mendoza is looking at about two years to cutover. This is well within the usual timescale, which Murray Smyth, IBS Group's senior vice-president for airline passenger services, puts at one to three years, end-to-end. But he warns: "Changes in leadership and business strategy - which are not infrequent - can radically change requirements and derail projects."
TAM chief information officer Juliana Kfouri agrees with this assessment. She is adamant that spending too long on a cutover is a risky strategy. "On average, airlines spend around two years understanding user needs, matching these with the system, and how one department connects to another, and how the business requirement is to be implemented. Then they change the chief information officer, then they change the president, then they scrap the project. I don't think you should spend five years on the change."
Driven by Kfouri's vision, a 700-strong combined TAM and Amadeus team took a whirlwind 12 months from TAM's decision to cutover to the end of the project. Communication was a key to success. "My objective at the beginning was to align internally, integrate with the users and define the scope very clearly with all the departments involved," explains Kfouri. "My charge was to integrate the users together with Amadeus to match their requirements. When we closed the scope and went to implementation, I had what we had to deliver to users very clearly in mind."
Clear communication extends to learning from other airlines. JetBlue had the benefit of following WestJet's cutover, and now LAN is talking to both to prepare for its project. "One of these airlines was very aggressive in risk-prevention. And, in the end, it paid off," says Mendoza. "Since the risk of these migrations is so very high, you have to be prepared to overinvest in risk-prevention and diminishing the impact of these problems."
JetBlue pumped $40 million into mitigating the risks of its transition to Sabre. The airline selected a Friday night in January, as this is a slower month, and Saturday is also quiet, as travel agents do not work weekends. In addition, it took 30 flights out of the schedule months in advance. "What was of more impact was we took the passenger loads down and capped flights at 40% load factors on the Saturday, and raised the cap over the next three days," says Rick Zeni, JetBlue vice-president for customer service system business change management.
Crucially, JetBlue also brought customers into the loop, warning of potential problems during cutover and the weeks that followed. This contrasts with WestJet's strategy, as the Canadian airline had concerns about competitors taking advantage of its cutover.
"WestJet's strong advice to us was to notify your customers," says Zeni. "Most of the lessons we learned [from WestJet] were positive, about what they did right. Most of what they did, they did very well. A couple of things they did caught them, and it's an illustration of how things can go wrong if a couple of things don't go right."
JetBlue built a new customised website and a back-up site, in case the primary one went down in the weeks following cutover. It also set up 400 temporary call-centre staff to handle run-of-the-mill enquiries, while the incumbent JetBlue team handled the more complex calls. On cutover weekend, it mobilised the leadership team and back-office staff into the airports, with IT people on standby to fix any unforeseen problems.
Zeni knew there might be one or two glitches. The trouble was, he did not know where. In the event, some kiosks and the website did not come up as quickly as required, and printers and scanners were not working at some airports. "The thing we probably underestimated was just how difficult it was going to be for our crew members to go from ticketless to the e-ticket system. It's really a mindset - the way our crew members interact with the system is much different, and it drove transaction times, which increased more than we thought," he says. The answer was to create groups to handle the more specialist transactions.
Such an intense level of risk-mitigation comes at a price, but for Zeni it was money well spent. "Clearly, we could have spent less money to do it a different way. But as I look back, I don't regret one dollar."
Source: Airline Business