What is a claims bordereau and how does it differ from a premium bordereau?

Where claims handling authority is delegated to a coverholder or third-party administrator, the insurer or managing agent still needs to know what is happening to claims arising under that business.

A premium bordereau cannot answer that question. It shows what was written, not what has since been claimed against.

That is why claims bordereaux exist as a distinct reporting document, alongside premium and risk bordereaux, wherever claims handling has been delegated.

Understanding what a claims bordereau contains, and how it differs from a premium bordereau, matters for anyone responsible for reserving, delegated claims oversight or monitoring customer outcomes under a binder.

What a claims bordereau records

A claims bordereau is a periodic schedule submitted by a coverholder or third-party administrator, listing claims notified, reserved, paid or closed under the delegated authority during the reporting period.

Typical fields include the claim reference, the associated policy or risk reference, cause of loss, date of loss and notification, current reserve, amounts paid to date, and claim status such as open, reopened or closed.

The emphasis is on the lifecycle of the claim itself: what has happened to it, and what financial position it currently represents.

How this differs from a premium bordereau

A premium bordereau is a transactional record of what was written and charged. A claims bordereau is a status record of what has happened after a policy was written and something went wrong.

The two are related, since a claim will usually correspond to an underlying risk that also appears on a risk bordereau, but they answer different questions. A premium bordereau answers "what business was written and what was charged". A claims bordereau answers "what has this business cost so far, and where does it currently stand".

Why the separation matters operationally

Reserving accuracy depends on claims-level detail, not premium totals. An insurer or managing agent needs an up-to-date view of open claims and current reserves to understand its financial exposure, which premium figures alone cannot provide.

Delegated claims oversight also depends on claims bordereaux to confirm that claims handled by a coverholder or TPA remain within the authority granted under the binder, including reserve authority limits and settlement authority.

Claims bordereaux additionally support monitoring of customer outcomes, since patterns in claim handling times, reserve movements or repeated reopening can indicate issues worth investigating.

How coverholders and TPAs traditionally report claims detail

In practice, claims bordereaux are often produced as spreadsheets structured around whatever the coverholder or TPA's claims system happens to export.

Status terminology in particular varies. One TPA's "Open" might correspond to another's "In Progress" or "Active", and a claim that has been "Reopened" might not always be flagged as distinct from a newly notified claim.

Tracking how a claim's reserve and status change from one reporting period to the next, across differently formatted submissions, has traditionally been a manual and time-consuming exercise for delegated claims oversight teams.

Where AI helps interpret and validate claims bordereaux

AI can help by recognising that differently labelled status values and fields represent the same underlying claim concept, and by tracking a given claim across successive bordereaux submissions to identify how its reserve and status have changed.

This reduces the manual effort of reconciling claims bordereaux that use different terminology or structures from one submission to the next, or across different TPAs and coverholders.

AI is well suited to flagging significant or unusual reserve movements for review. It is not well suited to deciding, on its own, whether a reserve movement is appropriate or whether a claim has been handled correctly. Those judgements remain with experienced claims and oversight professionals.

What to check before relying on claims bordereau data

Before treating claims bordereau data as a reliable basis for reserving or oversight reporting, delegated claims teams typically check completeness of submission against expected claim volumes, consistency of status terminology across periods, whether reserve movements are being tracked accurately from one submission to the next, and whether claim references reconcile with the corresponding risk or premium data.

Where any of these checks fail, the reported claims position should be treated as provisional until resolved.

Example

A Lloyd's managing agent delegates claims handling authority for a marine cargo account to a third-party administrator. Each month, the TPA submits a claims bordereau alongside the coverholder's premium bordereau.

The managing agent's claims oversight team uses AI to standardise claim status terminology across the TPA's bordereau and flag claims where reserves have moved significantly since the previous submission. Analysts review the flagged movements and confirm whether they reflect genuine developments or reporting inconsistencies before the reserving position is updated.

FAQs

• Does every delegated authority arrangement include a claims bordereau?

  Not necessarily. This depends on whether claims handling authority has been delegated under the binder. Where claims are handled centrally by the insurer or managing agent rather than delegated, a separate claims bordereau from the coverholder may not be required.

• How often are claims bordereaux typically submitted?

  Reporting frequency is set out in the binder agreement and commonly follows a monthly cycle similar to premium bordereaux, though this can vary by class of business and by the requirements of the managing agent or insurer.

• Can a claim appear on a claims bordereau before it appears on a risk bordereau?

  This is possible in principle if reporting cycles or systems are not perfectly aligned. Such a mismatch is typically treated as an exception requiring investigation rather than assumed to be an error.