I regard the operational usability of reinsurance as part of the purchasing decision, not as work to be discovered after renewal. A board paper can compare price, capacity and retention carefully while leaving uncertain whether the insurer can assemble the evidence required for a recovery. That uncertainty belongs in the same discussion. Buying protection and maintaining the means to use it are separate activities, but executives are accountable for the connection between them. A signed contract does not complete that responsibility.
I make this argument as an insurance operator, not as a substitute for specialist contract or actuarial advice. Technical interpretation matters, and a chief executive should know where their competence ends. My contribution is to ask whether the organisation's records, workflows and assigned responsibilities are consistent with the arrangement being proposed. If the answer depends on several teams taking action that has not been funded or assigned, the proposal contains an operating assumption that directors should be able to see.
Bring the recovery file into the renewal room
The claims colleagues who would prepare a recovery submission should encounter the proposed protection before the commercial decision has effectively been made. They can ask practical questions about the evidence available, the way losses are recorded and the distinctions the contract requires. Those questions may not change the preferred option. They may instead expose work needed to make it usable. Either result is more valuable before renewal than during the pressure of a widespread loss event, when missing records become harder to reconstruct.
At Brindleharbour Mutual, a sample of relevant historical claims found missing location-level evidence in 18% of reviewed files. We did not infer that every such file would produce a failed recovery. We did infer that our confidence in the process exceeded the evidence we had assembled. That was enough to require a management response. The sample created a concrete discussion between claims, finance and the reinsurance team, replacing a general assurance that the necessary data was held somewhere in the organisation.
The location record was only one dependency. We also examined who would distinguish losses associated with different events, who would reconcile submitted amounts and who would answer subsequent questions. Responsibilities that seem obvious in a diagram can become uncertain when people are working from different systems or definitions. I want those uncertainties recorded while they are still inexpensive to resolve. The purpose is not to burden the buying team with every operational detail, but to identify the details capable of changing the value of the purchase.
There is a timing problem. Renewal negotiations have deadlines, and data improvement rarely fits neatly within them. Management may be tempted to postpone the operating review until the next cycle. I prefer an explicit separation: what can be corrected before renewal, what will remain incomplete and what that means for the options available now. A temporary limitation can be acceptable if directors understand it and management owns the remedy. An unspoken dependency is much harder to govern because nobody has agreed the risk of carrying it.
Price the remaining operating dependency
Our 2024 decision combined record improvements with a protection structure better matched to the exposures we intended to retain. The chosen arrangement cost A$4.2 million more annually. I took that increase to the board with the work still required, rather than implying that the extra premium purchased freedom from operational difficulty. It reduced funds available for a planned distribution enhancement. Directors deserved to see that competing use of resources because the protection decision was also an allocation decision within the enterprise.
The cheaper alternative was not automatically irresponsible. It carried a different combination of retained exposure, terms and operating demands. We needed to explain that combination clearly enough for directors to exercise judgement. I resist papers that make one option appear obviously prudent by describing its benefits in detail and reducing every alternative to cost. A credible recommendation should explain what is being given up, which uncertainty remains and why management believes the chosen arrangement fits the institution's actual capabilities.
This also changes how we assess data expenditure. Correcting location records may look like an administrative project until it is connected to a recovery requirement. The financial case should not depend on an invented certainty about the next catastrophe. It can instead describe the weakness observed, the obligations affected and the cost of maintaining an adequate record. Directors can then judge the expenditure as part of running the insurance business, rather than as an optional improvement whose value is difficult to prove through a short-term revenue return.
Accountability must continue after the purchase. If a team is responsible for correcting missing fields, it needs a defined completion standard and a method of sampling whether the correction holds. A count of records touched is not enough. We want to know whether the right evidence can be retrieved and interpreted by someone who did not perform the original work. Otherwise the organisation may substitute a completed project report for a capability that still depends on a few experienced people remembering where material is stored.
Rehearse the evidence rather than the presentation
We used a synthetic catastrophe scenario to test the preparation of a recovery submission. The exercise did not seek a reassuring presentation about the quality of our arrangements. Participants had to assemble records, reconcile differences and identify missing material under a defined sequence of events. Two assumed hand-offs proved unclear. That was useful evidence because it identified a responsibility gap before a real event imposed commercial and human pressures that a controlled exercise cannot fully reproduce.
The subsequent file sample showed missing location-level evidence in 4% of reviewed cases. We reported the remaining exceptions and their causes rather than rounding the story into completion. Some required changes to how information was captured at inception; others concerned the transfer of material between systems. Those are different remedies. A single data-quality programme can conceal that distinction and leave recurring causes untouched. I want the operating owner to explain how new records will avoid the same problem, not only how historical files were repaired.
An exercise has limits. It cannot establish how a counterparty will interpret every future event or how a loss will develop. It can establish whether our own responsibilities are assigned, our evidence is retrievable and our internal account is coherent. Those are worthwhile findings if we describe them accurately. Treating a rehearsal as proof of a smooth future recovery would undermine the care that made it useful. Directors should see the uncertainty that remains alongside the weaknesses that have been corrected.
Before approving the next renewal, I would ask the executive team to bring one plausible recovery file into the room. Follow it from the underlying location and policy record through loss evidence to the proposed submission. Identify who owns each unresolved point and which part depends on specialist interpretation. The exercise need not displace the financial analysis. It should sit beside it, ensuring that the institution understands the operating work attached to the protection it is asking the board to buy.