How Jetstar is tapping data analytics

Even a 21-year-old company like Jetstar has legacy technology, the low-cost airline’s head of data, analytics and automation Alex Hopkins admitted at a Snowflake event in Melbourne.

As far back as 2019, Jetstar began a proof of concept with a view to using the Snowflake data cloud to replace an old, on-premise system that was nearing its end of life.

But then Covid-19 happened, throwing a major spanner into the works of the entire travel industry for a couple of years.

Things bounced back. Record-high levels of travel were seen in 2022, but “it was really tough” for airlines as they struggled with issues such as rehiring their workforce, resulting in lots of negative publicity around cancellations and other problems, Hopkins said.

By 2023, Jetstar’s data platform had reached its use-by date, so the airline signed a contract with Snowflake. The cloud consumption model helped this decision, Hopkins said, as the airline wasn’t sure how quickly it could move onto the new system.

The first use went live in 2024, and his team is currently moving more data assets onto the platform, and its vision of “safe self-service” for analytics is starting to materialise. Part of that is the creation of laboratory environments that allow employees to explore what they can do with the platform without any risk of affecting Jetstar’s live systems.

Hopkins outlined two applications the airline is running on the Snowflake platform. One concerns in-flight catering. Any perishable items that are loaded onto a plane but not sold go to waste, but having insufficient stock on board means disappointed customers and lost sales. The aim is to balance those criteria and identify the optimum quantities for each item and flight.

The other is for generating a schedule for Jetstar’s approximately 100 airliners. Obviously, each flight must be allocated a plane, but each plane must be subject to various types of maintenance within specific time windows. Consequently, there are a massive number of possible schedules, and producing a viable schedule has been a skilled manual task. Jetstar is investigating ways to automatically optimise these schedules, and to recover more smoothly when things go wrong, whether that is severe weather closing an airport, or a plane needing unscheduled maintenance before it can carry passengers again.

Hopkins pointed out that Jetstar’s on-time arrival record is better than Virgin Australia’s, and only slightly worse than its parent company Qantas’s, even though it spends less to achieve this level of performance.

Jetstar is in the process of adding more data assets to Snowflake, and the data, analytics and automation team is focusing on delivering as many high-value uses as it can. But there is still a lot to be done to get all the data they want into the hands of end-users. “We’re going strong,” said Hopkins.

The push for the adoption of artificial intelligence (AI) has come from business stakeholders, which is the opposite of the usual situation when new technologies arrive. At an airline, “safety is number one,” he said, so those stakeholders must be educated about the risks involved. “But we can’t be blockers,” he warned his peers, as that risks people running shadow AI projects that fall short of an organisation’s governance requirements. Business leaders need to understand the technology and realise the need for good data and guardrails before they can do interesting things, he added.

Hopkins’ plan for the next 12 months is to keep doing what they are doing: driving investment in the basics and looking for ways to unlock the value of Jetstar’s data.