Buyer hub

Cedent reporting AI: bordereaux out that match the treaty

Reinsurers reject or query files that do not map to wording: attachment, hours clause, excluded classes, currency, treaty year. The job is a file they can post, with a list of rows you still need to chase internally.

Reinsurers reject or query files that do not map to wording: attachment, hours clause, excluded classes, currency, treaty year. The job is a file they can post, with a list of rows you still need to chase internally. That is cedent reporting, not a transformation programme.

Inbox is the wrong metaphor here. The work is outbound. Premium, claims, and commission schedules leave the building on a calendar the treaty already named. Late is not a vibe. Late changes who owns IBNR and whether a reinstatement is on this year or the next.

The job

A cedent (or an MGA reporting as one) owes the reinsurer a periodic account of what was written and what was paid or reserved, in the shape the contract asked for. People still build that from policy admin extracts, claims extracts, and a spreadsheet someone formatted in 2019. Column names drift. Totals do not foot. The treaty year on the tab does not match the wording. Someone sends it anyway because the date on the slip of paper said it was due.

The operations loop is: extract the source rows, validate them against the wording, pack a bordereaux the counterparty can post, then chase the internal gaps. Claims recoveries sit on the same loop. If a claims row cannot be tied to the layer and the period, follow-the-settlements will not save you from a query.

You are not trying to impress the reinsurer with a narrative. You are trying to make their technician not open a dispute. Bordereaux are the instrument. Source-grounded extraction is how a total stays a total you can defend: document, sheet, cell, or the admin extract line it came from. No span, no value. A guessed grand total is how month-end becomes a negotiation.

Counterparties live on solutions for reinsurers. Their inbound job is the mirror of yours. If you send a clean file, they post. If you send a fight, they query. Software that invents a mapping to look complete is on their side of the fight, against you, whether you bought it or they did.

The calendar is in the wording. Monthly, quarterly, as-at dates, currency, and the basis for the cut. A reporting team that treats the due date as the day they start the extract will always be late. Freeze an internal cut-off earlier. Chase missing policy and claims rows inside the building. Then send a file with a gap list if some rows still have no source. A late complete-looking file is worse than an on-time file with holes you named.

Transport can be mail, a portal, or a drop zone. If mail is the path, the grant for the reporting mailbox should be as narrow as the placing grant on the broker side: read-only if you are ingesting replies and queries, and a separate conversation if anything is going to send. This website does not take live bordereaux. Production files stay in the tenant.

Three files, three jobs

Do not flatten premium, claims, and commission into one tab because the email subject said bordereaux. They are three jobs. See the difference between premium and claims bordereaux.

Premium bordereaux list what was written or ceded in the period: policy or certificate references, insureds, periods, limits, premiums, commission where it belongs on that file, class, territory. The completeness test is whether the reinsurer can see that the cession matches the treaty definition of subject business. A row with no class when the treaty excludes a class is a query. A premium total that does not foot to the lines is a query.

Claims bordereaux list losses: notice dates, loss dates, paid, reserved, recoveries, status. The completeness test is whether each row can be tested against hours clause, attachment, occurrence versus aggregate, and excluded causes. A paid figure with no loss date is not a claims row. It is a number.

Commission files (when separate) are how profit commission, sliding scale, and ceding commission get argued. Flattening commission into premium without keeping the basis is how a sliding scale gets applied to the wrong denominator.

Practical rules:

The bordereaux automation guide is the long form of ingest and mapping. Use it when the pain is format drift across many capacity providers. Use this hub when the pain is the cedent job itself: what leaves, against which wording, with what evidence.

Fictional discipline check: if ACME Construction Ltd (acme.example) appears on a facultative certificate, that risk's premium and any loss belong on the files that match that contract, not on a treaty bordereaux for a different class because the spreadsheet tab was handy.

Premium files fail in ordinary ways. Class blank. Territory blank. Period that overlaps two treaty years with no split. Gross and ceded mixed in one amount. A total row that is a typed number, not a sum. Each of those is a chase item or a query. Claims files fail when loss date, notice date, and paid date are one column, or when a reserve is a comment in the margin. Commission files fail when the sliding-scale denominator is not the premium the treaty named. Three jobs. Three tests. One email subject is not a reason to merge them.

Validate against the wording

Attachment, hours clause, territory, exclusions, class, currency, reinstatement. A total that does not have a source span is not ready to send. Validation is not a spellcheck on the cedent name.

Walk the file against the contract you actually signed, not the one you remember. If the treaty is excess of 5 million, a claims row reserved at 1 million should not appear as a recovery due. If the hours clause is 72 hours and the file has no event dates, you cannot apply the clause. If the territory excludes a region and the schedule has a location there, that premium should not be in the subject business without a conversation.

Validation outputs should look like the placing chase list, pointed inward:

The Bordereaux Agent is the named worker for turning messy premium, claims, and commission files into rows that map to treaty terms, with gaps instead of guessed totals. It does not settle as you. It does not tell the reinsurer you are right. It shows the span and the break.

If your current process is a technician reading the wording once a year and then mapping columns by memory, you will pass a file that looked like last year's. Treaties change at renewal. Hours clauses get negotiated. Clean-cut versus losses occurring is not a footer you can ignore because the template is pretty.

Reporting lag and clean-cut

Late bordereaux are not just an ops nuisance. They change who owns IBNR and reinstatement. Name the basis on the pack cover. See What is IBNR in reinsurance? when the argument is about incurred but not reported, not about whether your email was late.

Clean-cut (or similar cut-off language) and losses-occurring / claims-made style bases are not interchangeable. If the contract says the reinsurer leaves at year end on a clean-cut, a claims bordereaux sent six weeks late is not a courtesy problem. It is a who-owns-the-tail problem. If the basis is losses occurring, a premium file that omits in-force that attached in the period is a subject-business problem.

Lag also hides reinstatement. A catastrophe that sits in an unsent claims file does not stop the clock on whether a reinstatement premium was due. Sending a combined catch-up file without event dates is how both sides invent a timeline.

Operator habits that reduce lag fights:

  1. Print the treaty year, the basis, and the as-at on every file cover before anyone opens sheet 2.
  2. Freeze an internal cut-off that is earlier than the contractual due date, so chase happens inside the building.
  3. Do not wait for a perfect extract. Send a file with a gap list rather than a late file with invented rows.
  4. When a restatement is needed, label it a restatement with the prior as-at, not a quiet overwrite.

A gap list on time beats a complete-looking file that moved numbers to close the quarter. Reinsurers can work a gap. They cannot work a surprise restatement that pretends it was always that way.

Clean-cut language is easy to summarise and hard to operate. If the reinsurer is off risk at year end for unearned and for outstanding, a claims row that arrives after the cut with a loss date inside the year is not a casual addendum. It is a tail question. If the basis is losses occurring, omitting in-force that attached in the period understates subject premium and will be found when someone else foots the book. Print the basis on the cover so the technician posting the file does not have to open the wording to guess which universe they are in.

Restatements happen. Label them. Prior as-at, new as-at, what changed, which rows. A silent overwrite of last quarter's file is how follow-the-settlements evidence dies. The query six months later will ask what you sent, not what you now wish you had sent. Keep both.

Audit and follow-the-settlements evidence

Follow the settlements is not a magic phrase that closes every query. It asks whether the cedent settled in an honest and businesslike way, inside the contract. When the reinsurer asks why a claim was paid, you will need the loss file, the bordereaux row, and the wording clause you think responds — with sources, not with a recollection.

Keep the evidence pack as you go, not at audit time:

If extraction software filled a paid amount without a cell, you will not be able to follow anything except the model. That is why gaps are kinder than guesses. A missing paid figure is a chase. A guessed paid figure is a misrepresentation waiting for a commutation, a coverholder review, or a following-market audit.

We will not invent certificates or savings on this page. The test is whether you can open the span for a queried row in front of someone who does not work for you. If you cannot, the file was never ready.

Follow-the-settlements disputes are document fights. The reinsurer is asking whether the settlement sat inside the contract. Your answer is the clause, the loss file, and the bordereaux row, each with a source. A narrative from the claims handler is useful context. It is not the exhibit. If extraction filled paid without a cell, you will be defending a model, not a settlement. That is a worse position than a gap you chased.

Coverholder and MGA reporting adds a second audience: the capacity provider who wants the same files the reinsurer wants, sometimes on a different calendar. Do not maintain two undocumented mappings of the same extract. Name the contract on each file. If ACME Construction Ltd is facultative outwards and also sits in a binder, it is two lines on two files, not one convenience tab.

Agents and further reading

Implementation should look like a tenant drop zone or a scoped mailbox for the reporting team, one treaty wording as the validate-against target, and a first file that is allowed to have chase items. Read-only still applies if mail is the transport. This website does not take live client bordereaux.

If you also place facultative outwards, do not mix those packs into the treaty reporting inbox. ACME Construction Ltd on a facultative certificate is a different contract from the property quota share, even if the same technician owns both spreadsheets. Split the jobs. Name the wording on the cover. Send a file a reinsurer can post.

Questions

What files leave the building?
Premium, claims, and often commission bordereaux, plus supporting loss runs when the treaty asks. Each file should name the treaty, the period, and the basis (clean-cut versus losses occurring) you actually wrote. Rows need a source extract. A combined dump with no cover basis is how posting turns into a query. Do not flatten the three jobs into one tab to make the email simpler.
Why not send one spreadsheet with everything on it?
Premium, claims, and commission fail different wording tests. Class and territory belong on premium. Hours clause and attachment belong on claims. Sliding scale belongs on commission. Mixing them hides which test failed and invites a total nobody can foot. Keep three files or three bounded sheets, and put gaps on a chase list instead of inventing rows that make the tab look complete.
How does reporting lag change IBNR?
A late claims bordereaux is not only a missed due date. On a clean-cut, it can move who owns the tail. On any basis it can hide events that trigger reinstatement. Name the basis and the as-at on the cover. Send on time with a gap list rather than late with restated numbers and no label. IBNR arguments follow the file you actually sent, not the file you meant to send.
What evidence do we keep for follow-the-settlements queries?
The wording with page spans, the bordereaux as sent and as restated, the extract lines behind each row, and the query trail. Follow the settlements is not a substitute for a cited row. If a paid figure has no cell or document, you cannot show how you settled. Gaps are safer than guesses. Illustrations on this site are not a logo wall and not a substitute for that pack.