Customer challenge: Transform the finance department into the master of information for KLM’s business departments.
Provided solution: Develop and implement several use cases, including process mining for digital expense accounting, cost forecasting related to complaints and claims for delays and cancellations. Educate and train KLM employees on how to develop and productionize solutions.
Outcome: KLM’s finance department was able to independently continue developing and productionizing AI use cases.
The Dutch airline KLM wanted its finance department to become the master of information for its business departments. GoDataDriven developed several data analytics use cases and organized discovery workshops in Amsterdam and Paris to this end. KLM is now industrializing the resulting proofs of concept.
KLM’s Stefan van Heukelum clearly remembers the black suitcase. “To my surprise, it contained several nodes and a modem, which got the team started straight away,” he said. The GoDataDriven team had brought the suitcase over because KLM’s Hadoop big data cluster wasn’t available immediately after starting the project.
“Together we quickly built several pragmatic solutions, a few of which are now used in production. For me, this characterizes our cooperation with GoDataDriven,” said Van Heukelum.
Van Heukelum is one of the Dutch airline’s 30,000 employees and the product owner of data analytics within KLM’s finance department. He started working in this role in July 2017, the same year KLM transported 32.6 million passengers to 320 destinations in 114 countries. It was also the same summer he hired GoDataDriven’s data scientists, Steven Nooijen and Rogier van der Geer, to help him with financial data analytics and process mining at KLM.
The work of GoDataDriven has been of real added-value to our organization. They helped us combine operational data with financial data, and, thanks to that, we’re closer to becoming a master of information, said Stefan van Heukelum.
Big Data and Analytics
Jeroen Mulder works closely with Van Heukelum. As an operations-research analyst, Mulder was part of a team that led KLM to start working with big data and analytics a few years back and had very positive results. “Back then, GoDataDriven helped us build a recommender system for the commercial department and taught us a lot about data analytics. We really wanted to work with them again for this project,” Mulder explained.
“One of the most important things we learned about working with big data and external data scientists in the first few years was that the commitment from the business is of the utmost importance.”
“If people from different departments don’t feel a sense of urgency, nothing happens. They have to provide data or relevant issues for the data scientists, or you won’t go any further,” Mulder explained.
The Frontrunner of Big Data
KLM wants to be the most customer-centric, innovative, and efficient network carrier within Europe. Given the role of “big data frontrunner” within KLM’s finance department, Van Heukelum asked himself and his team, “How can we contribute to the success of KLM’s ambition?” The answer was to become a master of information, a partner to the business that can provide the smartest, most valuable information. “That’s why the department started focusing on data analytics and data-driven solutions – it’s necessary to add value and analyze processes thoroughly,” he said
Process mining for digital expense accounting was the first use case Steven Nooijen and Rogier van der Geer of GoDataDriven tackled.
“Accountants are familiar with sampling, but with data analytics, we can check all the data. We found that 90 percent of the processes went well,” explained Nooijen. “After that, we could dive deep into the ten percent of processes that weren’t carried out as instructed,” he explained.
The GoDataDriven team also worked on use cases for forecasting costs related to complaints and claims for delays and cancellations. These helped KLM determine how to best solve operational disruptions by looking at cost, as well as customer satisfaction. The team also used data science to improve online payment methods, initiate a performance-alerting system, and optimize the use of maintenance capital.
GoDataDriven’s proofs of concept were instrumental in KLM’s adoption of the data-driven way of working. One famous use case they presented to the CFO related to revenue leakage within the cargo department. KLM ships quite a few horses, which is very different from other shipments; the existing process wasn’t designed for it. “After combining several datasets, we found several discrepancies between the agreed price up front and what was eventually billed,” explained data scientist Van der Geer. “We realized a workaround was being used to organize horse shipments in KLM’s financial system, which created distortions in the revenue accounting process,” he explained.
Corporate controller David van Mechelen, who won the Leadership in Finance Award 2018, described this specific use case in a (Dutch) interview, underwriting the general adoption of data analytics within KLM’s finance department and the instrumental role of GoDataDriven in this transformation.
A Deep Dive into Numbers
“All change is challenging, especially in a traditional environment like finance and control,” explained Mulder. “At first, there was a bit of resistance from the finance workforce, but that quickly changed once people saw how the use cases could help them do a better job,” he said. “People who work in finance and control like to dive deep into their numbers, but the cyclical, demanding nature of their work often prevents them from doing so. Data-driven solutions can take over the repetitive work and help identify and define the issues they need to investigate thoroughly. Ultimately, data-driven solutions can make the work more challenging and interesting for our employees,” he explained.
In addition to building the proofs of concept, Nooijen from GoDataDriven also helped identify the pains and gains within the finance department over several discovery sessions using a value canvas. “The value canvas has proven to be a powerful tool for generating unconstrained ideas,” Van Heukelum remarked. “Consequently, we came up with four or five ideas that would add the most value. KLM has now embedded the value canvas into its ideation process.”
“In the ideation phase, the role of external data scientists is crucial,” explained Mulder. “They bring their experience of working with other clients to our table and combine that with the input from our employees, which results in extremely relevant use cases”.
Building and Developing Knowledge
The GoDataDriven consultants also helped build knowledge and understanding of internal data analytics at KLM. “One of our employees is a working student and what she learned from the GoDataDriven consultants was more valuable than a lot of what’s taught in the classroom. Other junior employees have also gained skills by working with Steven and Rogier,” remarked Van Heukelum.
“The GoDataDriven consultants are just as eager to learn as well. If something doesn’t work, they come up with new ideas until they find something that does. Every two weeks, they work a day at the GoDataDriven office, where they brainstorm with colleagues about the proofs of concept they’re building here. They always come back with a bunch of fresh ideas afterward,” Van Heukelum said.
Once KLM was able to develop the solutions further and productionize them, the GoDataDriven consultants moved on to start helping the next client.
Van Heukelum said, “We’re currently industrializing several of the proofs of concept Steven and Rogier built, so their work has been of real added-value to our organization. They helped us combine operational data with financial data, and, thanks to that, we’re closer to becoming a master of information. Our finance department has embedded data analytics into its daily routine, which has enabled us to become a better and more efficient organization.”
Project typeData Discovery Workshops
Proof of Concepts
“They bring their experience of working with other clients to our table and combine that with the input from our employees, which results in extremely relevant use cases.”