What a settled motor claim taught me about data

In the spring of 1997 I settled a rear-end collision on the outskirts of Liège for a policyholder who had done nothing wrong. We paid her garage within a fortnight and recovered the money from the other driver's insurer four months later. On my screen the file was finished and had cost us nothing. In the claims statistics it cost the full repair for one quarter and then disappeared. Nobody had made a mistake, yet the two records told different stories. The handler's record served the member; the statistical record served actuaries and finance colleagues who needed to know what motor claims really cost. I have spent most of my career on the distance between them.

I grew up in Liège with a Belgian father and a Danish mother, which is why my name puzzles people on both sides of the family. I studied economics at Hautmeuse University, graduating in 1995, and joined Lisaumont Assurances, the insurance subsidiary of a regional savings bank, as a motor claims handler that September. Handling claims gave me a respect for operational detail that analysts sometimes lack. A field left blank at five in the afternoon is usually blank because the handler had a garage on hold, not because of carelessness.

In 1999 Lisaumont moved me into claims information. Much of our home business was sold with mortgages in bank branches, and the bank's customer records rarely matched ours. I learnt to build reconciliations before building reports. By 2004 I managed a team of six producing the figures the claims director used every Monday, and I began working with the actuaries, who needed the same data for an entirely different purpose.

In 2009 my wife accepted a hospital post in Copenhagen, and I joined Ellekrone Forsikring, a Danish customer-owned non-life insurer, as its actuarial data manager. The chief actuary, Jens Ravnsholt, treated every claims triangle as an argument to be defended line by line. Ellekrone was preparing for Solvency II, and my job was to give the reserving team data they could trace to individual transactions rather than to annual extracts. I learnt professional Danish, completed a part-time certificate in actuarial statistics at the Lyngmarke Insurance Institute in 2011, and discovered that actuaries and claims handlers rarely disagree about facts. They disagree about which facts matter.

We returned to Belgium in 2014 when Arduvelle Assurances asked me to lead its claims and reserving data. Arduvelle was founded in 1926 by associations of craftsmen and small traders, and many members still run workshops, shops and building firms. That shapes the book. Income protection for self-employed people sits beside home and motor cover, and a single member's claim can touch their household and their livelihood at once. I led the reporting data for our first Solvency II year in 2016 and then the recoveries ledger that restated our motor reserve.

The chief executive asked me to become chief data officer in March 2019. The post was new, and I was clear from the start about what it would not include. Pricing stays with the chief actuary. Claims decisions stay with the claims director. Our data protection officer reports independently and can stop my projects. My office owns the platform, the definitions that connect one function's data to another's, the quality controls on data feeding reserves and regulatory returns, and the register of models that touch members. Since January 2025 the same inventory has also fed our DORA register of ICT third-party arrangements.

I lead through owners rather than through central control. Each of our 340 business terms has a named owner outside my team, and disputes about a definition go to the executive committee rather than being settled quietly in code. I present to the board's audit and risk committee twice a year on data quality in technical provisions and on the model register. I have not been a director, and I know that presenting to a committee is a different responsibility from sitting on one.

Within the Belgian market I am probably known for two habits: asking who owns a number before asking whether it is right, and publishing uncomfortable comparisons, such as the tenure gap in our home renewals, before anyone outside asks for them. Both habits come from that claims desk in Liège. A member experiences one insurer, whatever our organisation chart says. Our data should let us answer for that experience across claims, pricing and reserving, in plain language, to the member and to the board.

Open to Board appointments · Advisory roles

Languages / French · Dutch · Danish · English