From gut feeling to big data: this is how KLM continues to pioneer

At KLM, we work towards the future every day, and in the most innovative of ways. The phrase ‘It can't be done’ is not in our vocabulary, and Ali knows this all too well. As a data scientist, he designs models that improve operational processes within the company. Recently, his latest product went live: Chace. With this hand-luggage predictor, KLM takes another step towards a data-driven future.

Passengers in control

‘KLM has a very clear vision of how it wants to deal with its passengers,’ Ali begins. He paints a picture of the future. ‘Imagine: you book a flight, and on the day of departure your bags are collected at the front door. Passport control has also been done, so at the airport you walk straight to the gate. This won't be happening tomorrow, but it might be in 10 years' time. Bottom line: KLM aims to give passengers maximum control and ensure a seamless, surprise-free travel day. I help make that happen with data.’

Data points us to the way forward

Ali is helping KLM become the world's most data-driven airline. That goal was set five years ago. ‘In the first phase, we mainly looked at how we could improve large, critical processes. From aligning people and resources to optimising flight routes. We improved these operations by making them data driven. Now that this first improvement phase is behind us, we’re looking at less critical, but equally important processes. Hand baggage is one of them.’

Space shortage

‘Flights are often not completely full, and some passengers choose to travel without a suitcase. However, if every passenger brings a large piece of hand luggage, overhead space in the cabin quickly runs out. And that is a problem.'

From a gut feeling to data model

‘To avoid chaos at the gate, we aim to collect the right amount of hand luggage at check-in. The challenge is determining how much. Taking too much may cause frustration if there’s plenty of space onboard, while taking too little leads to stress and hassle during boarding.

We didn’t have an exact answer to the 'how much' question, so we relied on gut feeling. We guessed: Boeing 737 to Barcelona in May? Take in three pieces of hand luggage? To eliminate this guesswork, my colleagues and I developed a data-driven model that predicts the amount of hand luggage per flight. This improves the passenger experience and minimizes delays. The goal: no bag collection at the gate.’

Figures as greatest advisor

Ali has now been training the model for a year and a half, and a month ago we started using it in real life - initially for ten European flights. And with success. ‘The days of looking into the crystal ball are over. When we used to make vague forecasts, our gate agents often felt the storm brewing. Thanks to this new model, we can more accurately predict how much baggage needs to be taken to the hold. Much more accurately in fact: 85 per cent of the estimates are now correct. Before, it was only 40.

To make our data even more reliable, we need to improve its quality. We need commitment from the entire organisation for this. Which is quite a challenge because you don’t see the impact of better data straight away. KLM is taking this task very seriously. That's why we have had a department fully dedicated to data and AI for three years now. Fantastic, because that way we remain the most data-driven airline in the world.’