The question being asked is narrower than you think
People preparing a claim tend to imagine the review as a judgement on their honesty. It is much more mundane than that, and much more mechanical.
The person reviewing your file has to reach a conclusion they can defend to someone else. That conclusion has to rest on something other than your word, not because your word is doubted but because a file has to stand on its own once it leaves the room. So the working question is narrow: do these records support this figure, and do they agree with each other?
This reframing is worth internalising, because it changes what you spend your time on. Persuasive narrative does very little. A schedule where the invoices sum to the total, the total ties to the bank, and the bank ties to the accounts does almost everything.
It also explains a pattern that otherwise looks arbitrary: two claims of similar size and honesty, one settled in six weeks and one still open after eight months. The difference is rarely the merits. It is almost always how many rounds of questions the file needed before it could be signed off.
A necessary caveat
This article describes how claim files are typically reviewed. It is not insurance, legal or financial advice, and it cannot tell you anything about your policy — which is the document that actually decides what happens.
Policy wording governs everything
Cover, exclusions, indemnity periods and the basis of settlement are set by your policy. Two businesses with identical losses can have entirely different outcomes.
We do not assess or value claims
FlowParse reads documents into structured rows. What a loss is worth is an assessment, and it belongs to your broker, a loss assessor or your solicitor.
Practice varies
Adjusters, insurers and claim types differ. Treat what follows as the common pattern rather than a rule that will hold in every case.
Who the adjuster actually works for
A loss adjuster is appointed and paid by the insurer. That is not a criticism; it is simply the structure, and knowing it prevents two opposite mistakes.
The first mistake is treating the adjuster as your advisor — asking what you should claim, and taking a cautious answer as a limit. The second is treating them as an opponent, which produces a defensive file that answers questions grudgingly and takes twice as long.
The productive posture sits between: professional, complete, and quick to answer. An adjuster who receives what they asked for, in the form they asked for it, has no reason to look harder. An adjuster who receives three partial answers to one question has every reason to.
If you want somebody on your side of the table, that role exists and is called a loss assessor — you appoint them and you pay them. For a large or contested claim it is usually worth it. For a straightforward claim with good records it often is not, which is itself an argument for keeping good records.
The first three things that get checked
Reviews are not random. They start where an inconsistency would be most revealing, which in practice means starting with anything that can be compared against a source you did not produce.
| Check | Compared against | Why it comes first |
|---|---|---|
| Does the claim tie to the accounts | Filed accounts, management accounts | External, already stated, hard to revise |
| Does the trading pattern make sense | Prior years, same months | Seasonality exposes an implausible baseline instantly |
| Does the money move as described | Bank statements | Independent of anything you prepared |
All three have the same property: they compare your claim against a record that existed before the claim did. That is the whole logic. A figure that agrees with something written down before there was any reason to shade it is worth more than any amount of explanation.
The practical implication is that these three should be the first checks you run, before submitting. If your claim does not tie to your own accounts, you want to be the one who knows why.
Why two sources beat one perfect source
A single immaculate spreadsheet is less convincing than two ordinary records that happen to agree, and the reason is that a spreadsheet is something you made.
Invoices raised and bank receipts come from different directions. Till exports and banked takings do the same. A payroll record and a bank payment run do the same. When two such sources tell the same story, the story becomes hard to doubt without alleging something quite serious — and that is a much higher bar than ordinary scepticism.
They will not agree to the penny, and nobody expects them to. Timing differences, credit terms, floats, refunds netted off, a deposit banked the following Monday: all ordinary, all explainable in a sentence each. What matters is the shape agreeing and the differences being named.
The failure mode here is subtle and common: presenting only the source that reads best. The unused source does not disappear, and being asked why it was omitted is a much worse conversation than explaining a difference would have been.
What actually creates doubt
Doubt is rarely created by a large figure. It is created by small inconsistencies that have no obvious explanation, because each one raises the same question: what else has not been checked?
| What is seen | What it suggests | What removes it |
|---|---|---|
| Totals that do not sum | Nobody checked the arithmetic | Run the checks before submitting |
| Several suspiciously round figures | Estimates presented as facts | Label estimates and show the method |
| A period with no documents | Something is being left out | Disclose the gap and explain it |
| Documents dated after the request | Created for the claim | Send contemporaneous copies, note reissues |
| One supplier appearing three ways | Records are not controlled | Use the registration number as the key |
| Figures that change between versions | The number is being managed | One master schedule, versioned openly |
The last row deserves emphasis because it is the most damaging and the most avoidable. A figure that moves between submissions, even for a perfectly good reason, resets confidence in everything around it. If a number has to change, change it openly, say why, and reissue the whole schedule rather than a corrected extract.
Note also what is not in this table: being wrong. Errors are ordinary and expected. What creates doubt is errors that nobody appears to have looked for.
The round numbers problem
Real trading is untidy. Invoices end in odd pence, weeks have different numbers of working days, suppliers put through small credits. A schedule of exact figures looks like what it is: transcription from documents.
A column of clean hundreds reads differently, even when every figure is defensible. It suggests judgement was applied where measurement was possible, and the natural next question is how much judgement, and where else.
This does not mean estimates are forbidden. Some things genuinely cannot be measured — a fortnight of till data destroyed with the terminal, cash takings from a market stall, staff time split across two activities. Estimates are normal.
What matters is labelling. “Estimated: 2,400, being 12 days at the average daily take of the preceding eight weeks” is a defensible figure with a method attached. The same 2,400 sitting silently in a column of exact numbers is a problem waiting to be found — and when it is found, the question is not about that row.
The missing period, and why it is the worst thing to hide
Almost every claim file has a hole. A month where the bookkeeping fell behind, a system that was replaced, a box that went to landfill with the damaged stock. This is universal and not damning.
What determines the outcome is who identifies it. A gap you disclose is a fact about the world. A gap someone else discovers is a fact about your file — and it converts a review of the numbers into a review of you.
The disclosure itself can be short. What happened, over what dates, what was lost, what you have used instead, and how the reconstruction was built. Four sentences is usually enough, and it converts the weakest part of a file into evidence of care.
There is a second-order benefit worth knowing. Files that disclose their weaknesses tend to be sampled more lightly overall, because the reviewer has evidence that someone went looking. Files that present as flawless get tested until something is found, and something usually is.
Documents that arrive after the request
A document dated after the day it was asked for attracts attention, and it is worth understanding the distinction that is actually being drawn.
A contemporaneous document was created at the time of the event: the invoice the supplier sent in March, the delivery note signed on the day. A reconstructed document was produced later, often at your request: a duplicate invoice, a statement of account, a letter confirming what was supplied.
Reconstructed documents are perfectly legitimate and often necessary. The problem is only ever presentation. A duplicate invoice submitted without comment looks like an original until someone notices the issue date, and then it looks like something else entirely.
The fix costs one line: “duplicate obtained from the supplier on 14 June; the original was lost in the flood”. Same document, same figure, no question raised. This is the single cheapest credibility purchase available in a claim file.
Legibility is evidence quality, not housekeeping
A figure nobody can read is a figure that will be queried, and a receipt photographed at an angle in poor light is not a small inconvenience — it is a row somebody has to ask about.
This category is also the one with a deadline. Thermal receipts fade to blank within months. A supplier will usually reissue an invoice within a year or two and not after. Photographs of damage are impossible once a site is cleared, and clearing happens fast because trading matters more than paperwork in the week after a loss.
The practical rule is unglamorous: capture at the point of arrival, not at the point of need. Scan or photograph receipts the day they appear. Export from any system before it is decommissioned. Photograph the damage before the clean-up crew arrives, even when it feels like the least urgent thing in the building.
Where legibility is genuinely poor, flag it rather than transcribing a guess. A row marked as uncertain is a row that can be checked; a confident wrong number is one that gets found later, in the worst possible context.
There is a version of this that catches even well-organised businesses. A document can be perfectly legible and still be unusable as evidence because nobody can tell what it relates to — a delivery note with no reference, a card receipt with a merchant name that means nothing twelve months on, an invoice for “works as discussed”. The paper survived; the context did not.
The remedy is the same one that keeps coming up: a sentence written at the time. What this was for, what it replaced, who asked for it. It is the cheapest thing in the entire process and the only part that genuinely cannot be reconstructed, because the person who knew has moved on and the document never said.
What a good file actually looks like
Not thick. The instinct to submit everything is understandable and counterproductive: a file that takes three days to read gets read by sampling, and sampling is where inconsistencies get discovered.
One schedule, one row per document, with net, VAT and gross in separate columns.
A reference on every row pointing at the source file and page.
A short covering note: what happened, what is claimed, how it was built, what is estimated and why.
Estimates labelled as estimates, each with its method in one sentence.
Known gaps disclosed, with the reconstruction described.
Reconciliations shown: schedule to accounts, schedule to bank, invoices to receipts.
Documents supplied in the order of the schedule, not the order they were found.
The last point is the one people skip and the one that quietly saves a fortnight. Documents supplied in schedule order can be checked against the schedule directly. Documents supplied in the order they turned up require the reviewer to build an index first — and someone building an index is someone forming an impression of how the file was assembled.
The six questions, in the order they usually arrive
Claims differ, but the questions converge. Reading these before you submit is the cheapest review your file will get.
Can you show me how this figure was built? — the arithmetic trail from documents to total.
What would have happened anyway? — the part of the movement that is ordinary trading.
What did you save? — costs that stopped, which are always looked for.
Why was this cost necessary? — the reason behind an increased cost of working, not just the receipt.
Where is the document behind row 214? — asked to test the file, not that row.
Why does this not agree with your accounts? — timing, accruals and VAT basis, each answerable in a sentence.
The fifth is worth reading twice. Nobody cares about row 214. They care about how long it takes you to produce the document behind it, because the answer tells them how the whole file was built — and whether the rest of it can be relied on without checking every line.
What nobody expects you to have
Half the stress in a claim comes from imagining a standard that does not exist. It is worth being explicit about what is not expected.
Perfect records. Small businesses have gaps. Everyone knows this. What is expected is that you know where your gaps are.
Instant answers.“I will confirm that on Thursday” is a perfectly good response and far better than a fast answer that later changes.
Accounting fluency. Nobody expects a restaurant owner to explain accrual timing. They expect the underlying documents to exist and your accountant to be reachable.
A precise figure on day one. A range with a method behind it is normal early on. Precision arrives with the evidence.
Every original. Losses destroy documents; that is part of what happened. Duplicates and reconstructions are fine when they are described as such.
Why claims take months when they do
The elapsed time in a claim is mostly not work. It is waiting: a question is sent, it sits in an inbox belonging to someone rebuilding a business, an answer comes back partial, a follow-up goes out, and a fortnight has gone for each exchange.
Four rounds of questions is two months of calendar time and perhaps three days of actual effort. This is why front-loading matters so disproportionately: an hour spent pre-empting a question can remove a fortnight from the timetable, and the arithmetic holds for each question you anticipate.
| Stage | Well-prepared file | Assembled as questions arrive |
|---|---|---|
| First submission | Complete, with reconciliations | Partial, with a promise to follow |
| First response | Clarifications only | A list of missing items |
| Rounds of questions | One, sometimes two | Four or more |
| Time inside the business | Front-loaded, then finished | Spread over months, never finished |
| Effect on trading | A hard fortnight | A distraction all year |
The final row is the real cost and it never appears in any settlement figure. A claim that stays open occupies the person who runs the business, in the year when the business most needs them.
It is worth understanding why the rounds multiply, because it is not obstruction. A partial answer generates a follow-up almost mechanically: the reviewer cannot close the point, so they ask again, and the second question is usually broader than the first because they now have less confidence about what else might be incomplete. Two partial answers turn one open question into three.
If you are a small business without a finance team
Most of the advice written about claims assumes a finance function that produces monthly management accounts and can pull a supplier ledger on request. Plenty of businesses do not have that, and the advice still applies — it just has to be done differently.
What you almost certainly do have: bank statements, which are complete and independent; filed accounts, which are external and already stated; and a bookkeeping system or a shoebox with most of the documents in it. That is enough to build a defensible file.
The order that works when time is short. First, capture what is perishable — photographs, thermal receipts, anything on a system you might replace. Second, get the bank statements into rows, because they are the spine everything else hangs from. Third, match documents to the rows and note which rows have none. Fourth, write down what is missing and why.
That fourth step is what turns a modest file into a credible one. Nobody expects a six-person business to have a claims department. They expect someone to have looked, honestly, and to be able to say what they found.
The mechanical part — turning a year of PDF statements and a box of receipts into rows you can total — is described in insurance claim document processing, and the full sequence for a business interruption claim is in how to evidence a business interruption claim.
Four things people believe that are not true
“Submitting more helps.” Volume produces sampling, and sampling produces discoveries. A tight schedule with everything referenced beats a crate of paper in every case.
“Asking for less makes it easier.” Under-claiming does not buy goodwill. It produces a figure that does not match your own evidence, which raises exactly the question you were trying to avoid.
“They will not check the small items.” Small items are precisely what gets sampled, because they are quick to test and revealing when wrong.
“A good story helps.” Narrative explains context and cannot substitute for reconciliation. Where the two conflict, the reconciliation wins every time.
What to do today, if the loss is recent
In order, because the first three cannot be done later at any price:
Photograph everything before anything is cleared or replaced, with dates. This is the evidence that becomes permanently unavailable within days.
Capture the perishable paper. Scan or photograph thermal receipts now; they will be blank before the claim closes.
Export from any system you might replace — till, booking, stock. Once the hardware goes, so does the history.
Start a note of unusual spending with one line each on what it replaced and who authorised it. Thirty seconds now, impossible in a year.
Then, and only then, start on the bank statements and the baseline. Those records are not going anywhere, and they will still be there when the perishable evidence is not.