From the claim file to the reserving ledger

Mandate

Give claims, pricing, reserving and reporting one governed set of facts with named business owners, and bring every model that affects members under recorded oversight.

Outcomes

• Launched income-protection claims triage in 2023; median time to first payment fell from 19 days to 10
• Built a renewal comparison for 248,000 home policies that led the board to cap the tenure gap in January 2025
• Established a register of 23 models with owners, permitted decisions and monitoring by member group
• Cut late reporting adjustments after quarterly cut-off from 52 to 9

Decisions in context

I became Arduvelle's first chief data officer with a mandate the chief executive wrote in a single sentence: give claims, pricing, reserving and reporting one set of facts that someone owns. I spent the first year mapping who already owned what, and found that most of our disagreements concerned definitions rather than numbers. The data platform moved to a single claims and policy model by 2021, and the model register followed in 2022, when the analytics team began building its first claims models. The income-protection triage launch and the home renewal comparison are the decisions from this role that I describe in detail. Both required me to bring evidence to colleagues who owned the outcome and to accept that they, or the board, would decide. I sit on the executive committee and present twice a year to the audit and risk committee.

Reason for the move

Still in post; an audit or risk committee seat elsewhere would sit beside this role, not replace it.

Mandate

Make claims and reserving data fit for Solvency II and reconcile the claims system with actuarial and finance records.

Outcomes

• Delivered reporting data for the Solvency II opening balance and the first quarterly returns in 2016
• Rebuilt motor claims history from 2011 with recoveries as separate entries, adding €7.2 million to the motor best estimate
• Reduced the reserving team's manual adjustments from 41 to 6 a quarter

Decisions in context

Arduvelle hired me to make its claims data fit for Solvency II eighteen months before the regime applied. The first period went on reporting data for our opening balance and the first quarterly returns in 2016, which we delivered with more manual work than I would accept today. The second was the recoveries ledger. I learnt there that a technically sound fix fails without a settlement between the people who own the outcome: actuaries who feared losing an independent view, and a claims director whose handlers had to record more on every motor file. I reported to the chief financial officer, which gave the work authority during the close but sometimes led claims colleagues to see my team as finance's inspectors. When the chief data officer post was created, I asked for it to report to the chief executive for exactly that reason.

Reason for the move

The chief executive asked me to take the newly created data post on the executive committee.

Mandate

Give the reserving team claims data traceable to individual transactions and prepare data for Solvency II best-estimate calculations.

Outcomes

• Replaced annual claims extracts with a monthly transaction-level reserving database
• Documented 120 transformations between the claims system and the reserving triangles
• Supplied data for Solvency II best-estimate trial runs at the 2012 and 2013 year-ends

Decisions in context

Copenhagen began with some humility. My Danish was a childhood language, adequate for family dinners and inadequate for an actuarial meeting, and Ellekrone's chief actuary, Jens Ravnsholt, expected every number in a triangle to be defended. The company was a customer-owned mutual preparing for Solvency II, and its reserving team worked from annual extracts that nobody could trace to individual claims. My team of seven built a monthly reserving database at transaction level and documented each of the 120 transformations between the claims system and the triangles. Several turned out to be workarounds for problems fixed years earlier. I completed a part-time certificate in actuarial statistics in 2011, mainly so that I could follow the actuaries' arguments rather than only supply data for them. The lesson I carried home was that actuaries and handlers agree about facts more often than either expects.

Reason for the move

Returned to Belgium for family reasons and to lead claims and reserving data at Arduvelle.

Mandate

Produce the weekly claims figures used by the claims director and reconcile claims data with bank and finance records.

Outcomes

• Introduced a weekly claims pack with reconciled counts and payments
• Matched branch-sold home policies to bank mortgage records, resolving 14,000 mismatched addresses
• Began supplying the actuarial team with monthly claims extracts

Decisions in context

Running claims information at Lisaumont meant producing the pack the claims director read every Monday morning and defending it whenever a branch manager disputed it. Much of our home business was sold by the bank's branches alongside mortgages, and the bank's address records and ours disagreed for about 14,000 policies. That mattered when a storm hit one district and we could not say how many members lived there. My six analysts and I reconciled those records over eighteen months. I also began sending monthly claims extracts to the actuarial team and discovered how differently they read the same data. Where the claims director saw a backlog, the actuaries saw a change in settlement speed that would distort their projections. That double reading became the centre of my work, and I left only because my family moved to Denmark.

Reason for the move

Moved to Copenhagen when my wife accepted a hospital post there.

Mandate

Turn claims-system data into reports that branch and claims managers could rely on.

Outcomes

• Built the first monthly reconciliation between claims payments and the general ledger
• Separated recoveries from payments in motor claims reports
• Documented the claims coding used across the Liège and Ghent offices

Decisions in context

My first data job was an apprenticeship in reconciliation. The claims system, the finance ledger and the branch reports each produced a motor claims figure, and none of them matched. I spent most of 2000 building a monthly reconciliation of claims payments with the general ledger and learnt that most differences had ordinary causes: timing, a recovery recorded as a negative payment, a duplicated cheque cancelled in one system but not another. In 2001 I split recoveries out of the motor reports because branch managers were celebrating months in which other insurers happened to pay us back. I also documented the claims coding used in the Liège and Ghent offices, which differed more than anyone had admitted. I had no team, but I had the handlers' trust, because I had been one of them and asked before changing anything they recorded.

Reason for the move

Promoted to lead the claims management-information team.

Mandate

Settle motor claims fairly and promptly and pursue recoveries from other insurers.

Outcomes

• Handled about 900 motor claims a year
• Pursued recoveries from other insurers under the market's settlement arrangements
• Trained two new handlers in liability assessment

Decisions in context

I joined Lisaumont's Liège claims office straight from university and handled about 900 motor claims a year: own-damage repairs, liability disputes and, increasingly, recoveries from other insurers under the market's settlement arrangements. A team leader with thirty years on the desk taught me to read a police report and a garage estimate side by side and to telephone before I wrote. The job taught me what claim data is for. Every field on a file exists to help settle that case, and handlers complete it under time pressure with a member or a garage waiting. Years later, when I asked handlers to record more for reserving purposes, I remembered the cost of a field. I also learnt that a claim can be settled for the member while money is still owed to the company. That gap became the subject of my career.

Reason for the move

Moved into claims information when the company formed its first claims reporting team.