A bank reconciliation analyst works the cash. A bordereaux analyst works the file: every row of every monthly bordereau checked against the binder terms, the last version you accepted, and the risk, premium and claims files beside it. Commission, tax, premium movements and claims movements are proved first, so the cash match at the end has something true to match against.
| Clm Ref | Pol No | DOL | Paid | O/S Res |
|---|---|---|---|---|
| CL-8802 | HX-40219 | 3/5/26 | 4,200.00 | 12,000.00 |
| CL-8815 | HX-40221 | 11-May-2026 | 0.00 | 35,500.00 |
| CL-8820 | HX-40230 | 2026/05/19 | 18,750.00 | 0.00 |
| CL-8829 | HX-40236 | 27.5.26 | 2,310.40 | 6,000.00 |
| CL-8833 | HX-40240 | 02/06/2026 | 9,875.00 | 15,125.00 |
| CL-8841 | HX-40377 | 14/7/25 | 26,400.00 | 40,000.00 |
| Claim ref | Certificate | Date of loss | Paid in period | Outstanding |
|---|---|---|---|---|
| CL-8802 | HX-40219 | 2026-05-03 | 4,200.00 | 12,000.00 |
| CL-8815 | HX-40221 | 2026-05-11 | 0.00 | 35,500.00 |
| CL-8820 | HX-40230 | 2026-05-19 | 18,750.00 | 0.00 |
| CL-8829 | HX-40236 | 2026-05-27 | 2,310.40 | 6,000.00 |
| CL-8833 | HX-40240 | 2026-06-02 | 9,875.00 | 15,125.00 |
| CL-8841 | Held back for review, row left untouched | |||
147 of 148 claims clean and sent on in the carrier's format. 1 held with a reason; checked against May's file and the risk bordereau.
Bordereau rows processed, mapped, de-duplicated and checked
Discrepancies flagged for human review
From bordereau received to processed and checked
“Timely access to risk, premium and claims data is critical. Ensuring data adequacy will be a core focus of our oversight.”
“With only 43% of premium payments made on time, there is room for improvement across our market.”
“A faster bordereaux process is not sufficient to resolve what is effectively a control problem.”
The file says what was written. The bank says what arrived. Reconciling the two, with evidence, is the control. Who does what in bordereaux management →
Both get called reconciliation, and both end with a number that has to agree. They start from different documents, ask different questions and catch different mistakes.
| Bank reconciliation | Bordereaux reconciliation | |
|---|---|---|
| The question | Did the money arrive, and what does it belong to? | Is what the partner reported right, and has it stayed right? |
| The inputs | Bank statements, broker remittances, premium finance statements | This month’s premium, risk and claims bordereaux, the binder or treaty terms, and every prior version of the file |
| Checked against | What was bound: the policy, layer or treaty each receipt settles | The contract (commission, brokerage, tax, levies, share), the last version accepted, and the sibling bordereaux for the same program |
| What it catches | Short pays, unallocated receipts, duplicate and misdirected settlements, bound but unpaid | Commission drift, tax and levy errors, silent premium changes, disappearing rows, claims that moved backwards, written premium with no paid behind it |
| Cadence | Daily, as cash lands | Monthly, per partner, each file against the last, plus the look back over the history when a dispute or an audit opens |
| Who runs it | Credit control and cash application | Bordereaux management, delegated authority operations, reinsurance accounting |
| The Analyst | Reconciliation Analyst | Bordereaux Analyst, this page |
The two jobs meet at the end. Once the bordereau is proved, the Reconciliation Analyst matches its lines to the cash that actually arrived. Written against paid is the last check, not the first.
A bordereau is the monthly report a coverholder or MGA sends to the capacity provider that carries the risk it has written, and that a cedant sends up to its reinsurers. Under delegated authority it is the only record the carrier has of the business written in its name. Three files usually travel together, and a fourth follows once the money moves.
Risk bordereau
Every policy written in the period: insured, inception and expiry, class, limits, and the premium charged.
Premium bordereau
Every premium transaction: new business, endorsements, cancellations and returns. Each row starts at gross, then commission, brokerage, taxes and levies come off to the net due to capacity.
Claims bordereau
Every open and closed claim: paid to date, outstanding reserve, incurred, status, and the movement since last month.
Paid bordereau
The premium that was collected and remitted. It arrives a month or more behind the written file and rarely lines up with it row for row.
The carrier books premium, pays commission and reserves claims on a file it did not produce. Checking it is the control.
Commission, brokerage and tax come off before the money moves. A commission rate that drifts two points shows up as premium that quietly never arrives.
Commission and tax are set by class and by jurisdiction in the contract. The bordereau rarely says which rate it applied, so each row has to be tested against the terms.
Written against paid can only be reconciled across periods, with every prior version of the file intact. Lose the history and the gap is unprovable.
For the full chain, who sends what to whom and which system of record each party keeps, see where Brisc fits in the bordereaux chain.
Each one runs on every row of every file the day it lands. The cash match comes fourth.
The commission rate on the row against the contracted rate. Tax and levies against rate and base for the row’s jurisdiction. The sign against the transaction type. Duplicates, premium outliers, invalid dates and identifiers, currency and FX. Where the program’s footing is set up, gross less commission, brokerage, taxes and fees must equal the stated net.
Every check, named and explained →Every version of every row is kept, and each new file is diffed against the last one accepted. Premium that moved with no endorsement or cancellation behind it. Rows present last period and missing this one. On claims, a status that went backwards, incurred that fell without a recovery, recoveries netted instead of coded, a claim that vanished without closing.
The three files for a program checked against each other rather than read in isolation. The written bordereau against the paid bordereau: balances outstanding longer than the program allows, paid lines with no written line behind them, endorsements that do not tie back to their policy. No bank statement required. Runs as a recurring monthly audit across every program.
Once the file is proved, the Reconciliation Analyst matches each line to bank statements and broker remittances, with the evidence attached. A readiness check on the bordereau flags any row missing the fields the cash match will need, a month before it would have failed. On the cash step, 80% of records match on day one, rising to 92 to 95% by about 90 days.
Matching the cash is only defensible if the line items underneath it reconcile. The Analyst tests every bordereau row against the terms before anything downstream uses it.
The expensive errors are quiet. They show up only when this month’s file is laid beside the last one you accepted, row by row.
The real risk isn’t inefficiency. It’s the error that was small in the month it happened. The commission rate that drifted two points in month four. The policy that dropped out of the file with no cancellation behind it. The claim that reopened without anyone seeing the reserve move. Each compounds until the audit, the dispute or the claim finds it.
Every check is code against a named rule, so a flag is reproducible and a total is computed, never generated. AI reads the documents and suggests how a new partner layout maps to your schema; a person confirms the mapping before it is used.
A partner’s layout is mapped once and reused every month. When their template changes, safe changes are applied and noted; the rest go through a guided update. The file is not halted.
One library of checks, toggled and tuned for each program’s terms. Enforcement is off by default: every flag queues for a person, nothing is corrected silently.
Every version of every row is retained with who acknowledged what and when. The history is there when a dispute opens, and month-on-month checks always have a baseline.
Accuracy you can defend. Output accuracy holds at 97%+ on the extracted and checked file, with the evidence stored on every flag. On the cash step, the Reconciliation Analyst matches 80% of records on day one, rising to 92 to 95% by about 90 days.
Every rate on every line tested against the binder and the jurisdiction, the day the file lands.
Each file diffed against the last accepted version; premium that moved and rows that vanished are named with the prior row beside them.
Reported premium checked against what settled, across periods, without waiting for year-end.
A reinsurer replayed four years of cedant bordereaux through the same checks and found a 12% uplift in true profitability.
Every flag carries its evidence: a number an auditor, a capacity provider or a board can follow back to the row.
The process that works at ten programs still works at fifty, without a hire per increment of volume.
“The beauty of Brisc was taking about 80% of the manual labour away, so people can focus on underwriting instead of data entry.”
| Manual process | With the Bordereaux Analyst | |
|---|---|---|
| Ingestion | Every coverholder’s layout re-keyed into working spreadsheets | Any partner layout, mapped once and reused every month; template changes handled without halting the file |
| Commission and tax | Checked on a sample, when there’s time | Every row: rate against the binder, tax against rate and base per jurisdiction, footing set up per program |
| Month on month | This month eyeballed against last month, if last month can be found | Every version kept and diffed against the last accepted; silent changes and disappearing rows named |
| Across files | Risk, premium and claims read separately, by different people | Checked against each other and against the paid bordereau, as a recurring monthly audit |
| Written vs paid | A year-end exercise, reconstructed from old files | Running, with the proved lines handed to the Reconciliation Analyst for the cash match |
| Exceptions | Buried in a “to investigate” tab that nobody reopens | Queued with the row, the check and the reason; acknowledged with who and when |
| Audit | Reconstructed at year-end from emails and old spreadsheets | Every version, every flag and every decision kept, exportable, current every day |
| Growth | Another analyst for every few programs added | The same process at ten programs and fifty: capacity without headcount |
For MGAs & coverholders
Commission, tax and footing proved before the bordereau leaves, and the paid bordereau built from matched cash rather than rebuilt by hand.
Where Brisc fits for an MGA →For insurers & carriers
Coverholder bordereaux checked on arrival against binder terms and last month’s file, written against paid running, aged debt surfaced while it can still be chased.
Where Brisc fits for a carrier →For reinsurers
Cedant statements checked to treaty and layer, gross-up and commission conventions applied per treaty, and the history re-run when a dispute opens.
Where Brisc fits for a reinsurer →The Analyst reads what your coverholders, cedants, brokers and banks already send. No new templates, no format mandates.
Bordereaux, statements and remittances in whatever shape they arrive.
WillisHowden+ regional brokers+ your coverholdersMost arrive as Excel. Market-standard and not-remotely-standard alike.
XLSXCSVPDFLloyd's 5.2Email attachmentsStatements ingested directly from the accounts you already run.
HSBCLloydsBarclays+ yoursUploads in, checked file and exception queue out: run in parallel with your existing process, then graduate to posting into your PAS, general ledger or data lake once the numbers have earned it. Deployment: 2 to 6 weeks.
Dedicated database, storage and key vault on Azure. No shared fallback.
Entra SSO federation with an allowlist you control.
SaaS. No infrastructure for your team to stand up or maintain.
SOC 2 Type II certified, with per-tenant isolation. Security documentation: trust.brisc.ai.
Looking for cash application against your bank feeds, bank statement first? See Bank Reconciliation.
Cyber SME, embedded, micro-policies: a monthly file with thousands of lines, each worth too little to check by hand. Sampling catches the pattern and misses the row. Checking every row is the only way these books get reconciled at all.
Bordereaux reconciliation is checking the premium, risk and claims bordereaux your delegated authority partners send against three things: the contract (commission, brokerage, tax and levy rates, shares), the prior versions of the same file, and the other bordereaux for the same program, including the paid bordereau. Done monthly and line by line, it proves that commission, tax, premium movements and claims movements are right before the figures are booked, reported up to capacity or settled. Matching the lines to cash is the last step, and it only means something once the file itself has been proved.
Bank reconciliation asks whether the money arrived and what it belongs to: bank statements and remittances matched to bound policies. Bordereaux reconciliation asks whether what the partner reported is right, and whether it has stayed right: the file checked against the binder, against last month’s file, and against its sibling files. One job works the cash; the other works the file. Brisc runs both. The Bordereaux Analyst proves the file and the Reconciliation Analyst matches the cash, and the written-against-paid check sits where the two meet.
Yes, on every row rather than a sample. The commission rate on each line is checked against the contracted rate, and tax and levies are tested against rate and base for the row’s jurisdiction. Where a program’s footing is set up, gross less commission, brokerage, taxes and fees must equal the stated net, and a row that does not foot is flagged with the discrepancy named. Nothing is corrected silently: every flag queues with the row, the check and the reason for your team to decide.
Yes. Every version of every row is kept, and each new file is diffed against the last version you accepted. Premium that moved with no endorsement or cancellation behind it, rows that were present last period and are missing this one, and claims that moved backwards or lost value without a recovery are each flagged with the prior row beside the new one. Because the history is kept, it can be re-run: one reinsurer put four years of cedant bordereaux, 100,000+ premium records, through the same checks and found a 12% uplift in true profitability.
Yes. The risk, premium and claims files for a program are checked against each other rather than read in isolation, and the written bordereau is checked against the paid bordereau: balances outstanding longer than the program allows, paid lines with no written line behind them, and endorsements that do not tie back to the policy they adjust. It runs as a recurring monthly audit across every program.
No, deliberately. Brisc does not initiate, hold or move a payment. Once the file is proved and the cash is matched, the Analyst tells you what arrived, what it belongs to, what is outstanding and what to remit to each capacity provider, and hands you the outputs ready for the banking rails you already use. That is also why deployment takes 2 to 6 weeks: adopting Brisc is not a treasury change or a new regulated dependency.
Every check is deterministic code against a named rule, so a flag is reproducible and a total is computed, never generated. AI appears in two places: reading the documents, and suggesting how a new partner layout maps to your schema, which a person confirms before it is used. Output accuracy holds at 97%+, with the evidence stored on every flag. On the cash step, 80% of records match on day one, rising to 92 to 95% by about 90 days as the rule base absorbs your brokers, aliases and netting conventions.
2 to 6 weeks. Day one needs no PAS integration: this month’s and last month’s bordereaux go in as they are, and the checked file, the version diff and the exception queue come back in the formats you already work with. You run the Analyst in parallel with your existing process until the numbers earn trust, then graduate to posting into your PAS, general ledger or data lake.
Both. The Bordereaux Analyst reads source data in whatever form it arrives, a partner file, a PAS extract or a feed from your data warehouse, and generates the bordereau in the format the receiving party requires. Once the cash is matched, the premium bordereau you report up to capacity is generated from validated line items, so the file you send is built from the reconciliation rather than rebuilt alongside it.
30 minutes, no slide deck. You watch the Bordereaux Analyst read both files as sent, run the checks, show what changed between them, and hand the proved lines to the cash match.
Book the walkthroughNot ready for a call? Get the MGA operations report →