How Do We Manage Version Control of Mapping Rules?

Bordereaux mapping is never a one-off exercise.

Coverholders change their reporting templates, insurers update target schemas, new classes of business are added and regulatory requirements evolve. Every one of these changes can require an adjustment to the mapping rules that translate a coverholder's bordereaux into an insurer's standard format.

If those changes are not properly recorded, organisations quickly lose track of which mapping rule applied to which submission. That makes it difficult to explain historic data discrepancies, reconcile figures during an audit, or defend a position during a claims dispute.

Version control of mapping rules is the discipline of recording what changed, why it changed, who approved it and when it took effect, so that any historic bordereaux can always be traced back to the rules that produced it.

Why mapping rules change so often

Mapping rules are rarely static because the things they depend on are not static either.

Coverholders periodically restructure their bordereaux, adding, renaming or removing fields, often to reflect changes in their own systems or local regulatory requirements. Insurers update their target templates as reporting standards evolve or as new classes of business are added to a binder. Regulators themselves introduce new reporting obligations that require additional data points to be captured and mapped.

Each of these events can require a change to the mapping logic. A coverholder adding two new fields for a local regulatory requirement, for example, may mean a new mapping rule needs to be created from scratch, while an existing field might need to be remapped if its meaning has shifted.

Because these changes happen continuously and often independently of each other, version control cannot be treated as a one-off task completed when a mapping is first built. It has to be an ongoing operational discipline.

Traditional approaches to controlling mapping rule changes

Most delegated authority organisations have historically managed this through some combination of the following.

Spreadsheet-based change logs. A shared spreadsheet or document records each change to a mapping rule, typically including the date, the reason, and who made the change. This is simple to set up but relies heavily on manual discipline to keep up to date.

Version-numbered mapping documents. Each significant revision of a mapping specification is saved as a new numbered version, with older versions retained for reference. This preserves a history but can become difficult to search or reconcile against specific bordereaux submissions.

Manual sign-off and approval workflows. Before a mapping change goes live, it is reviewed and approved by someone with oversight authority, often via email or a change request form. This introduces accountability but adds administrative overhead, particularly when changes are frequent.

These approaches work, but they scale poorly. An organisation managing mapping rules for dozens or hundreds of coverholders, each subject to periodic changes, can find itself with an unwieldy patchwork of spreadsheets, document versions and email trails that are hard to search when a discrepancy needs investigating months later.

Where AI helps maintain traceability

AI-assisted mapping tools change what is practical here, not by removing the need for governance, but by reducing the manual burden of maintaining a reliable audit trail.

When a coverholder's bordereaux format changes, AI can detect the shift, for example recognising that new columns have appeared or that an existing field's content no longer matches its previous mapping, and propose an updated mapping rule. Crucially, it can log exactly what has changed compared to the previous version, rather than requiring someone to manually document the difference.

This means every proposed change carries its own record: what the old rule was, what the new rule is, and what triggered the change. That record can be retained automatically alongside an effective date, so that a bordereau submitted in March can still be reconciled using the mapping rules that were live at the time, even if the rules have since been updated.

What AI does not do is decide, on its own authority, that a change should go live. The proposed update still needs to be reviewed and approved by someone with oversight responsibility. AI reduces the repetitive work of detecting and documenting change; it does not replace the judgement involved in approving it.

Operational considerations for governing mapping rule changes

A few practical points matter regardless of which tools are used.

Every mapping rule change should be traceable to a business reason, not just a technical description of what was altered. Recording that a field was remapped because a coverholder changed its template is far more useful during an audit than a note that simply says a column moved.

Historic bordereaux must remain interpretable using the mapping rules that were in force when they were submitted. If mapping rules are updated in place without preserving prior versions, an organisation loses the ability to accurately reprocess or reconcile older submissions.

Approval should sit with someone holding oversight authority, distinct from whoever makes the technical change. This separation is not bureaucracy for its own sake. It ensures that mapping changes are assessed for their operational and reporting impact, not just their technical correctness.

Finally, change control should be built into routine operations rather than treated as an exceptional event. Given how frequently coverholder formats and target schemas shift, a lightweight but consistent process will hold up far better over time than an ad hoc one invoked only occasionally.

Example

A Lloyd's managing agent receives monthly bordereaux from a marine cargo coverholder. The coverholder updates its reporting template to add two new fields required by a change in local regulation.

Without a controlled process, the operations team could adjust the mapping rules informally, leaving no record of why the change was made or when it took effect. Months later, this would make it difficult to explain why certain historic bordereaux look different from more recent ones.

Instead, the delegated authority operations analyst logs the coverholder's template change, and an AI-assisted mapping tool proposes an updated mapping rule for the two new fields, flagging exactly what has changed compared to the previous version. The oversight manager reviews and approves the update before it goes live. The tool retains both the old and new mapping versions with an effective date, so any historic bordereaux can still be reconciled using the rules that applied at the time it was submitted.

FAQs

• What happens if we don't version control our mapping rules?

  Without version control, organisations risk being unable to explain historic data discrepancies, struggling during audits, and applying inconsistent treatment to similar bordereaux submitted at different times. Reconstructing why a figure looks different from one month to the next becomes guesswork rather than a documented fact.

• Who should be responsible for approving mapping rule changes?

  Approval should sit with someone holding oversight authority, separate from whoever makes the technical change. This separation ensures changes are assessed for their operational and reporting impact, not just technical correctness, and is a matter of good governance rather than unnecessary bureaucracy.

• Can AI decide on its own when a mapping rule should change?

  No. AI can detect changes in source formats and suggest updated mapping rules, along with a clear record of what has changed, but the decision to adopt that change should remain with a delegated authority professional. Human oversight stays in the approval loop.