Author: Tessa van de Kamp

Translating a high-over ambition into a concrete implementation. That was the main challenge for Anna Geraedts, Data Science Manager of Arbo Unie. Together with Anton Bubberman of Anderson MacGyver, they built a solid data foundation that should contribute to the vitality and health of working people and thus make organizations more successful.

Anna Geraedts, who hails from the occupational health and safety sector, was asked by Executive Board member Willem van Rhenen to help Arbo Unie do more with data and digitalization based on its strategy. The organization was still in the pioneering phase in this area. “That gave me the opportunity to build everything from scratch,” Geraedts says. “In doing so, I was able to use all my previously accumulated knowledge and experience. For example, how to put together a data-science team and avoid getting stuck in BI and reporting.”

If you really want such a team to come up with predictive models, for example, you have to clearly ask for them. “Then they can deliver very concrete answers and solutions. For example, in the area of planning or predicting behavior,” says Geraedts – not a data scientist herself, but as a PhD psychologist and through experience in similar roles eminently data-savvy. She has worked at Soulve Innovations, HumanTotalCare, and ArboNed, among others.

Within Arbo Unie, it quickly became clear that much more was needed than simply rolling out the data warehouse acquired shortly before she joined the company. In the interaction with Anderson MacGyver, it was decided to take a programmatic approach, with a clear structure and objectives. Not just building, but designing a data organization and laying down a data foundation was the route.

Issues
“For the first few months in particular, I sought input from all levels, including the Executive Board,” says Anna Geraedts. “What are the main pain points, where do people see new opportunities and where do we ultimately want to go as an organization? We then relatively quickly outlined the program together with the people from Anderson MacGyver.”

This involved several issues, according to Anton Bubberman, management consultant and Guild lead Data at the consulting firm. “What issues regarding data management and data architecture needed to be internally managed and what could we manage externally? But also the question of what business roles you need, for example, to improve data quality and to establish ownership of the data.”

Geraedts: “Gradually it became clear what the possibilities are for extracting value from data and what had to happen in terms of processes and systems in conjunction for this. In the process we ran into various challenges, for example that some customers or individual users could be located in different ways in the systems.”

“It was and often is also difficult to be able to unlock certain data from crucial legacy systems,” adds Bubberman. “There was also the privacy issue surrounding the proper handling of personal data and compliance with legislation. Medical privacy in particular is very sensitive.” All such issues were addressed together with techies, business people and management.

Five streams
The data program within Arbo Unie was divided into five streams: realizing data awareness, governance and organization, the data foundation, platform & data product realization, and materializing the potential value by defining, prioritizing and planning the data initiatives.

A number of important steps have since been taken. “We have established a data warehouse that can be used for the medium term. In addition, in a tremendously tight labor market, we managed to recruit about a dozen data professionals within a quarter. The backbone is now in place and we can now really get to work with data.”

What helps, according to Bubberman, is that Arbo Unie really has a pupose, a socially relevant mission, to which the data strategy and data specialists contribute significantly. Anna Geraedts: “It’s true that we are often chosen over that axis. It’s nice to be able to mean something for the Dutch labor market through your work. What we do really has an impact on people’s lives.”

Sparring partners
According to the Data Science Manager, the value of Anderson MacGyver lies in its broad involvement. “Several consultants with diverse knowledge and experience as true sparring partners got the organization and the Executive Board on board. Anton’s commitment was fundamental in getting everything around data management off the ground, such as the data warehouse and data architecture.”

“Several consultants acted as real sparring partners in getting the organization along“ Anna Geraedts – Data Science Manager at Arbo Unie

Bubberman: “Some of those involved in the program focused mainly on the organization, while Anna and I were more on the data and technology side. The great thing about Anna is that there is really someone standing there, taking steps and getting things moving.” Given the nature of Arbo Unie, that is sometimes quite a task. “An organization where people want to deliver quality, to do things really well. That’s very logical in a field where it’s about people’s health, but it sometimes slows things down.”

Challenges
Arbo Unie still faces challenges, according to the Data Science Manager: “What will the organization look like going forward, which people do we bring in and how can we retain them?” Anton Bubberman adds: “To that end, we still need to get a better view of what we really want to achieve as part of the digital strategy.”

Around the data foundation, moreover, there are technical aspects at play that have not yet been tackled. “We are struggling, as I said, to extract data from some crucial systems and new data sources also need to be added, in order to get new use cases based on that. Once that is solved you automatically get a snowball effect,” said the management consultant.

Starting with external data is an option, although he says the focus is first on getting their own data in order. “If we can then combine that data with data from other sources, a lot is indeed possible, and exploratory work can be done quickly. But scalable data science solutions in production are challenging.”

When asked for a tip or recommendation, Anton Bubberman mentions the importance of a looming perspective at all levels, in which the coherence of all steps and initiatives is also clear to everyone. Anna Geraedts agrees: “Make sure you get the organization on board at the managerial level, so you can always make the right decisions quickly.”