A Bermuda-based specialty reinsurer used Brisc’s Bordereaux Analyst to audit four years of premium and claims bordereaux, more than 100,000 premium records. The audit found a 12% uplift in true profitability hidden in misallocated and unprocessed data.
A multi-year program had closed looking profitable, and the routine controls had signed it off every quarter. Leadership still wanted to know whether the profit was real and the controls were holding; standard reporting could answer neither.
Losses paid against records no bound policy explains.
Dates validated against the policy, not the spreadsheet.
Rows that pass a monthly check and fail a four-year one.
Collateral movements no bordereau accounts for.
Four years of bank statements, premium and loss bordereaux, and treaty documentation, normalised into one reconcilable dataset.
Deterministic, explainable matching cross-references policies, claims and payments against the collateral accounts; anomalies are flagged, never smoothed over.
A complete audit trail on every finding, treaty-level breakdowns, and a categorised discrepancy register with resolution recommendations.
The discrepancies summed to a 12% uplift hidden in misallocated and unprocessed data, with every finding evidenced back to the source row, because the Analyst read everything instead of sampling it.
Errors that individual monthly bordereaux pass through cleanly but that compound across a program: claims with no matching policy, losses booked outside policy periods, premium anomalies, and unexplained trust account variances. Each row looks plausible; the pattern only shows when every row is reconciled.
Yes. This engagement was a back-book audit: four years of premium and claims bordereaux, bank statements and treaty documentation, more than 100,000 premium records, reconciled end to end.
Matching is deterministic, rule-based and explainable. AI reads the documents and argues the borderline cases; it never decides a match. That is what makes the findings defensible to auditors and counterparties.
A complete audit trail on every finding, treaty-level performance breakdowns, and a categorised register of data quality issues, each with a resolution recommendation.
A sample can validate rows; it cannot see patterns between them. An orphaned claim or a drifting trust balance only surfaces when every policy, claim and payment is reconciled against the others, which is exactly the work manual review cannot afford to do.
You cannot manage risks you cannot see, and you cannot see them by sampling once a year. Bring a real bordereau; we run it and you keep the findings.
Book the walkthrough