The month closes whether you've checked it or not
Every coverholder writing under one of your binders owes you a return, and the cycle does not pause for your year-end, your system migration or your staffing.
What arrives is whatever their system could export. Somebody on your side opens it, works out which column is which, and checks the business written against what the binder allowed. Then the next one lands. SnapLine reads those returns as they come in, reconciles them against the binder, and hands back what does not agree.
The format is whatever the coverholder's system produces
There is a data standard for delegated reporting in this market. Whether a given coverholder can produce it depends on the system they run, and plenty of them run something that was chosen before the standard mattered to anybody.
A new binder adds a format. Nothing ever removes one.
What that looks like in practice:
- Written risk, premium and claims arriving as one workbook, as three files, or as three files on three different days
- Column headings that describe the coverholder's own system rather than the binder
- Spreadsheets, delimited files, and returns produced as PDF by somebody who only had a print button
- Cumulative figures where last month was incremental, or the reverse, with nothing on the file to say which
- Currency and date conventions that differ by coverholder and occasionally within a single file
- Risk and peril coding that has to be mapped before anything can be summed
- Locations given at whatever granularity the underlying system happened to hold
Nothing here asks the coverholder to change what they send. Mapping a new format is set-up work done once for that coverholder, rather than reading work done again every month for as long as the binder runs.
Checked against the binder
A reconciliation that compares this month against last month tells you the return is internally consistent. It does not tell you whether the business inside it was within the authority you granted.
The binder is the reference. Classes, territories, periods of attachment, limits, deductibles, exclusions and the premium basis are terms, and a return either agrees with them or it does not. SnapLine holds those terms and tests the return against them.
What surfaces is usually unglamorous. A risk written in a class the binder does not name. An inception date outside the period. A limit above what the binder delegates. A premium total that does not equal the sum of the risks underneath it. None of that is interesting on its own. It is interesting because it is already in your systems and nobody has looked at it.
Where a line cannot be reconciled because something is missing rather than wrong, it is marked as missing and becomes a query back to the coverholder. The gap is the finding, not an obstacle to producing one.
What arrived this month
- Thornby Underwritingmonthly_return_mar.xlsxPOL_REFINC_DTGWP_GBPTIVPERIL_CDOne workbook, three tabs. Risk, premium and claims.
- Marrowgate Coverthree files, three daysPolicy NoAttachingPremium to dateSum InsuredPremium reported cumulatively. Nothing on the file says so.
- Calderhall BindingMarch bordereau.pdf— no column headings —Printed from their system. Fourteen pages, no grid underneath it.
- ClassesAs scheduled in the binder
- TerritoryUnited Kingdom and Republic of Ireland
- Period of attachment01 Apr 2026 to 31 Mar 2027
- Limit any one risk5,000,000
- Premium basisGross, as written
What did not reconcile
- Fails a binder termExposureLimit above what the binder delegatesAgainstLimit any one risk, 5,000,000Found7,500,000CellThornby · monthly_return_mar.xlsx · tab Risk · row 412 · column H
- Fails a binder termExposureClass not named in the binderAgainstClasses as scheduledFoundEngineering — plantCellMarrowgate · risks_2603.csv · row 103 · column B
- Fails a binder termExposureInception outside the period of attachmentAgainstPeriod, 01 Apr 2026 to 31 Mar 2027Found12 Mar 2026CellMarrowgate · risks_2603.csv · row 88 · column C
- Fails a binder termPremiumPremium total does not equal the sum of the risks underneath itAgainstPremium basis, gross as writtenFoundHeader total over the lines by 11,480CellThornby · monthly_return_mar.xlsx · tab Premium · cell F2
- Missing, not wrongExposureLocation given to postcode district onlyAgainstNothing failed. The value is not there to testFoundQuery back to the coverholderCellCalderhall · March bordereau.pdf · page 6 · third block
What happens to each if nobody reads the return
Money reconciles against money. The figure meets something downstream that contradicts it, and it gets found whether or not anybody opened the file.
A wrong sum insured, a wrong class or a location in the wrong zone has no such property. Nothing downstream disagrees with it. It sits in the aggregate being correct-looking until something forces the question.
Where the numbers go next
The return is where your booked premium comes from. It is also where your aggregate exposure comes from, and your claims movement. Whatever is wrong in the file is wrong in your systems from the day it is loaded, and it stays wrong until something forces the question.
Premium errors do eventually get found, because money reconciles against money. Exposure has no such property. If a coverholder has been writing into a zone your aggregate already sits heavily in, the only thing between you and learning that at renewal, or after the event, is how quickly the return got read and understood.
Downstream of all this sits reporting you are accountable for, where the figures are expected to tie back to a source. SnapLine does not prepare that reporting and makes no claim to. What it changes is the quality of what the reporting is assembled from.
The hours spent reconciling by hand are the visible cost, and they are the smaller one. Aggregation you cannot see is the part that actually costs money.
Exceptions, with the cell they came from
The output is a list. Each item says what failed, which binder term it failed against, and where in the coverholder's file the value sits.
That last part is the discipline the rest of the product runs on. A field extracted from a submission links back to the region of the document it was read from, and a bordereau line that fails a check links back to the row and the column it failed on. An exception nobody can trace is an exception somebody re-derives by hand before they can act, which puts the work back where it started.
An item can be cleared, corrected, or sent back to the coverholder as a query, and whichever happens is recorded against the line. Next month opens on a position somebody can read instead of on an email thread.
Oversight is its own job
Most of this site is written for an underwriter deciding whether to take a risk. Your job is a different one. You are accountable for business somebody else has already bound, under terms you agreed and cannot police as they are being used, and the evidence available to you is a file that turns up once a month.
That changes what is useful. An exception list that reads back cleanly into a binder review, a performance discussion or an audit is worth more to you than processing speed, because the question you get asked is never how quickly the return was read. It is what you did about what was in it.
Send us last month's returns
Pick the coverholder whose file nobody volunteers to open. The one with the merged header row, or the one who puts three months of movement in a column labelled adjustments.
If what comes back is an exception list you would be prepared to take into a binder review, the next step is a fixed-scope Proof of Value on your own live returns, scoped in writing before it starts and ending in a written report.
Any measurement of what this is worth happens there, on your book. We are not going to put ours on a web page.