“I’m an engineer at heart,” says Stuart Birrell. “So this is fun.”
Fun, for the chief data and information officer at EasyJet, means running one of the biggest e-commerce operations in the UK, at the same time as overseeing the IT behind the world’s eighth-largest airline.
“You’ve got the real high energy, high speed, high cadence of an e-commerce, digital environment. But then you also have a very safety-conscious, highly regulated airline operation, with 15,000 crew and 30 bases, multilingual, having to run the day-to-day operations, subject to weather, regulations, all the rest,” he says.
“Bringing all that together in one organisation is a huge challenge. How to put systems and data and people in place who can operate that breadth of complexity and mindset and demand is hugely challenging.”
It sounds exhausting. “I love the job,” says Birrell. “Despite the inevitable frustrations.” But his are frustrations that many an IT leader would be equally enthused to take on.
Birrell has built a career around working in high-profile, high-performance environments. Before moving into IT, he helped build the Nissan factory in Sunderland, and the Walkers Crisps factory for PepsiCo. In between stints as CIO at first Gatwick Airport and later Heathrow, he spent four years as CIO at supercar maker McLaren Group, owner of the Formula One team.
With that in mind, perhaps it’s no surprise that, in November 2020, he moved to an airline right in the middle of the sector’s pandemic-induced crisis. Even that turned out to be an opportunity.
“When I got here, there was a very clear mission from the CEO and the board. We’d come through the pandemic, there were ups and downs, but there was a recognition that we needed to rebuild the company, technically and capability-wise,” he tells Computer Weekly.
“The pandemic showed that we had very good systems, but running the airline during the disruption of the pandemic showed we didn’t have the flexibility and agility we needed. What had become very tuned, very optimised and very efficient wasn’t going to be sufficient for a more agile, flexible world looking forward. The challenge I was given was, how do we rebuild this place?”
The three years since have been focused on improvements in two major areas. First, modernising the key commercial systems – that is, the front-end EasyJet website, the booking engine behind it, and the airline inventory supporting it. Second, was to move completely away from in-house datacentres to the cloud.
“During the pandemic, there were about 1,200 new airlines started up around the world, which prompted the question: if we were starting again, how would we do it? And building datacentres wouldn’t be part of that. So let’s go cloud-native,” says Birrell.
“Two-and-a-half years from when we said ‘go’, by the end of this year we will be out of datacentres. We’ll be rid of the big old networks, using SD-WAN [software-defined wide-area networks], zero-trust [security] and 100% cloud-based. And that is really paying some big dividends already.”
The technology transformation is allowing the business to think differently, he says, opening up new opportunities and provoking the sort of cultural and mindset shift that comes from a growing understanding of the benefits of integrating data and democratising access to it across the organisation.
“When you give people good data, they can make things sing and dance,” he says. “New capabilities, new ways, new opportunities, that the commercial teams and operations teams can really start to exploit. Now we have the platforms and foundations – my god, the opportunity there is huge.”
“Being in the cloud is worth 10 times more than going to the cloud. The mindset, the ability, the agility, the capability it gives you as a company is way more beneficial than whatever you save from a hosting contract”
Stuart Birrell, EasyJet
For Birrell, this is the real justification for becoming a cloud-native business. It’s not about saving money – even if exiting datacentres will cut costs. It’s about what you can do once you’re in the cloud.
“Being in the cloud is worth 10 times more than going to the cloud. The mindset, the ability, the agility, the capability it gives you as a company is way more beneficial than whatever you save from a hosting contract or whatever. That’s the beauty of it – getting people to understand what it gives you. That’s the real power,” he says.
EasyJet is a multicloud company – not so much through a conscious decision, more as the inevitable outcome of choosing the right environment for each application.
“Every company is dependent on all three big cloud suppliers, whether you know it or not,” he says. “If you look down through your supply chain, you’ll realise that what you think of as software as a service is actually one of the three biggest providers.”
This was amply demonstrated in the UK summer heatwave of 2022, when Google’s London datacentre overheated, disrupting services for one of EasyJet’s third-party suppliers despite the airline having no direct contractual relationship with Google.
“The complexity of supply chains is a very real problem. As long as you’re in the cloud, you’re never going to get away from that.”
Similarly, the IT infrastructure is a mix of in-house resources and external partnerships. The software stack combines packaged applications with bespoke developments, making widespread use of application programming interfaces (APIs) for integration. Outsourcing giant TCS is a key partner, involved with cloud operations, service desk and other project work.
“I’ve got partners and consultants and whoever doing all of the doing, the graft work,” says Birrell. “But it’s a team internally leading it and the accountability sits with my team.”
The booking engine is the heart of the company’s e-commerce operation – an aircraft is filled with passengers through the EasyJet website and mobile app every 10 seconds. The complexity that sits behind your choices when booking a flight is immense.
“For each of those bookings, you’ll get maybe 200 searches into the database. That means billions of searches and calls on the database for inventory availability. We make 1,200 calls a second on some of these APIs. It’s about keeping that going 24/7, with a very low tolerance for failure,” he says.
“When you look at our revenue of around £8bn this year, most of that is coming through the mobile app and website, through our booking engine. We can’t afford downtime.”
The benefits from the work on data integration are really coming into play. Birrell’s department includes a small team of data scientists, but they are part of a community of about 120 data scientists and analysts at all levels across the organisation – “everything from PhD data scientists down to spreadsheets”.
“I don’t need to run all the data scientists to get the value of the data. There’s a framework, I’ve got the data platform, the data quality and controls, but you know what? Get people to solve business problems and opportunities and you don’t have to ask IT. So it’s about democratising and getting it out there. And that change in mindset and desire and capability is fascinating to watch,” says Birrell.
Stuart Birrell, EasyJet
“We do a lot of predictive analysis, a lot of planning, simulations, and we are embarking on a full digital twin of running the fleet,” he says.
EasyJet has more than 320 aircraft, flying from 30 bases to around 150 destinations. “That’s a big optimisation. There are constraints on landing slots and access time. We have three sizes of aircraft, but which aircraft goes on what route? What time of the day will maximise customer experience, service, and all the rest of it? So, how do you bring all that together into a network plan with a flight schedule and a crew list? We’re stepping that up to the next generation of sophistication and quality to bring it together, so you can have one view [of the entire fleet].”
AI and machine learning (ML) already support that operational planning because, as Birrell points out, “the permutations are just mind-boggling”. His team has so far developed 18 AI and ML algorithms which are live in production. He’s also looking into what opportunities quantum computing could bring to further optimise planning, although he acknowledges that the technology is “not quite there yet, but it will get there”.
Birrell has been an enthusiastic early adopter of generative AI (GenAI), which is already delivering significant benefits in EasyJet’s contact centre, in streamlining financial reporting for its holidays business, and in querying the extensive manuals used by flight crew (see, How EasyJet is using generative AI, below).
As with his approach to multicloud, Birrell follows a “horses for courses” principle when it comes to the GenAI models in use.
“We made a conscious decision to be model agnostic. The tools we built, we’ve done them with the API such that we can go to Claude or we can go to ChatGPT, or whichever ones are coming up next. It’s moving so rapidly you can’t be reliant on just one tool,” he says.
“We’re finding from the testing we’re doing that different models have different strengths. Some are really good with email, some are good on query generation. Everything has strengths and weaknesses, so we can choose. And as you bring in new models, it’s an API call to switch over. So we are using ChatGPT, we are using Claude, and the Google tool as well. We have all three of them live and that will continue to evolve.”
In contact centres
In October 2023, the company went live in contact centres with generative AI (GenAI) analysing incoming emails. For example, if a customer’s flight has been delayed, they might send an email with a claim or a query about compensation. Where it used to take an analyst 40 minutes to read, understand and compose a response to the email – using GenAI, that’s down to about 10 to 15 minutes.
The GenAI application produces a summary of the inbound email, proposes a response and puts that in front of the contact centre analyst so there’s still a human in the loop. The quality of the answers is so good that the number of follow-up questions has dropped by 30%.
Querying crew manuals
Take a scenario where one of EasyJet’s aircraft based in, say, Geneva is delayed in Lisbon. What are the rights of the passengers? What must the crew do? What is needed in terms of overnight lodgings and other requirements? The answers historically lie in manuals which can be several inches thick. Every one of EasyJet’s 30 bases will have different rules.
The manuals have been put into a GenAI tool, so staff can now use a natural language query to find out what the crews are entitled to and the passenger support available.
Internal private ChatGPT for staff
EasyJet has implemented its own private ChatGPT for internal staff. “The productivity of people using that already is just incredible,” says Birrell, citing two examples.
The EasyJet holidays business, which is autonomous and runs separately for legal reasons, uses the tool to generate its own financial reports. Birrell’s IT team has written only two structured reports for the whole billion-pound holidays business. Finance and commercial teams use natural language queries to generate reports. That capability is now rolling out across the airline. “We’ve eliminated the whole data cube, all the structured reports and all that stuff,” says Birrell. “The things you’ve historically needed coders to do, you don’t need that anymore.”
The HR department ran an internal staff survey and received a couple of thousand comments back. Instead of manually reading through and analysing all those comments, the team put the survey into the private ChatGPT and produced a “very good” summary of the issues in 15 minutes.
Generative AI is not without its controversies in the tech market. For every expert who extols the potential benefits, there’s someone else warning of the risks and, in particular, the threat to jobs. As an early adopter, with production systems already in place, what advice can Birrell offer to fellow IT leaders looking into the technology?
“There are two things. One is the confidence to just do it – the technical confidence. All it is, is an API call. Technically, it’s not a data science issue, it’s an engineering problem,” he says.
“Second – and the biggest one – is governance. Because we’ve been working on AI for a while, we have a good multifunctional governance group looking at this. You have to show you’ve got the guide rails so that you’re not putting data into an unsafe environment. Design that in from the front.
“We’ve had a lot of challenging conversations and you have to take people with you, because they don’t understand it. Most people in IT still don’t understand how large language models work, for example. And you have to understand how they work – why it’s safe, and how the safeguards work in the different models, because AWS has a different approach to Microsoft. The technical side was easy. The contractual and governance side, data privacy, all of that is the single biggest issue.”
The experience of putting GenAI applications into the hands of staff has been hugely positive. Birrell says people can see the benefits and the opportunities because it’s helping them to do their jobs more effectively.
“It changes people’s roles and allows them to go and do the thinking around how to solve problems, rather than spending half their day going through everything and getting bored by the end of it. It’s shifting how people want to do things. My biggest challenge has been managing the demand and the pull – I’m not pushing this and persuading people to use it, which is a great place to be,” he says.
In October 2023, EasyJet announced an ambitious expansion programme that will bring a further 157 aircraft into the fleet over seven years – with an option for a further 100. Birrell says the experience gained from using AI and optimisation is going to be key to enabling that growth.
“How do you take a successful airline, and take all that optimising and AI, to enable us to grow significantly by 20% or 30% in volume? We’ll have more passengers, more aircraft, more staff, more crew. How do you optimise that? The next bit is about how we start optimising and automating the planning and scheduling in the airline operations space,” he says.
“The whole operational management is the next big investment area and focus. And that’s shuffling the team to pivot to that space and to take the big simulations and digital twins to understand how you run an airline efficiently and economically using these tools.”
The “huge challenge” that Birrell says is so much fun rests on EasyJet’s ability to use technology and data to get 100 million passengers to their destination every year, on time, with a great experience.
“[At EasyJet] there’s a recognition of the role that technology plays, and the role of IT is changing as well. It’s impossible for IT to manage all the technology and all the data. What you have to do is lead it and set direction and build capabilities and let the company grow around you,” he says.
“That’s why it’s one of the best jobs in the world.”
Read more interviews with CIOs in high-performance environments
- Stuart Hughes, chief digital information officer, Rolls-Royce: The aerospace giant uses internet of things and sensor technology to create a personalised digital twin for the airlines that fly its engines, but the benefits of further data science and analytics could be even greater.
- Clare Lansley, CIO, Aston Martin Formula One: A rare female CIO in a male-dominated sport, Lansley discusses how digital transformation is all a part of helping the team to move further up the F1 grid.
- Marc Jennings, CIO of analytics and AI, TUI: The international travel group is overhauling its data stack and aims to provide self-service analytics to key employees across the organisation.