An asymmetric conversation
On one side there is an organisation that processes claims all day, with forms, checklists and people whose entire job this is. On the other there is you, doing it for the first time, in the weeks after something went wrong.
That asymmetry is not anyone’s fault and it is not going away. What it means practically is that the burden of assembling a coherent account falls on the side least equipped to carry it — and that the quality of that account has a large effect on how long everything takes.
The documents almost always exist. Invoices, receipts, statements, orders: they are in email, in folders, in a drawer. What is missing is the bridge between having them and being able to hand over something that can be checked.
This page is about that bridge, and about where a tool that reads documents stops — because on this subject the line between help and overreach matters more than usual.
What this is not
Stated first rather than in a footnote. Insurance is a field where the wrong kind of confidence causes real harm, and it is better to know the boundary now.
| Question | FlowParse | Who does it |
|---|---|---|
| Tell you what your policy covers | No | Your broker or a solicitor |
| Value a loss | No | A loss assessor or accountant |
| Assess or settle a claim | No | The insurer and its adjuster |
| Negotiate on your behalf | No | A broker or loss assessor |
| Submit anything to an insurer | No | You, or whoever represents you |
| Read documents into structured rows | Yes | — |
| Check that the arithmetic holds | Yes | — |
| Keep every figure linked to its file | Yes | — |
Nothing on this site is insurance advice. The whole contribution is mechanical: taking documents you already have and turning them into rows that add up and point back at their source.
That sounds modest and it is where a surprising share of the delay actually lives.
What a claim needs, in document terms
Policies and claim types differ enormously, and underneath them the document requirements rhyme.
| What is asked | What satisfies it | Where it goes wrong |
|---|---|---|
| What you had | Invoices, asset records, photographs | Bought years ago, paperwork gone |
| What it was worth | Purchase documents, valuations | Only a memory of the price |
| What you spent because of it | Invoices and receipts after the event | Mixed in with ordinary spending |
| What you lost | Accounts, statements, order records | No comparable period to point at |
| That you paid | Bank statements | Kept apart from the documents |
| When it happened | Dated records | Dates recalled rather than recorded |
The third row is the one that catches people out. Costs incurred because of the loss have to be separable from costs you would have had anyway, and nothing about a supplier invoice announces which it is. That distinction has to be made by someone, at the time, or reconstructed later at much greater cost.
The fifth row is a mechanical failure and the easiest to prevent: the payment proves money moved and the invoice says what it bought. Filed in different places, they take twice as long to present and invite a query for no reason.
Two different things are being proved
Keeping them apart makes the whole exercise easier to organise, because they need different material and different people.
That the event happened, and is covered. Reports, photographs, correspondence, the policy itself. This is largely narrative and legal, and it is where a broker or solicitor earns their fee. Nothing on this page helps with it.
What it cost. Numbers, backed by documents, that a stranger can add up and check. This is bookkeeping under pressure, and it is entirely mechanical.
Almost all avoidable delay lives in the second. The first tends to be settled by facts that are either there or not; the second turns into weeks of correspondence when the figures cannot be traced to anything.
Practical consequence: separate the two piles from the start. Trying to build one document that does both produces something that does neither well, and it is the most common shape of a claim submission that comes back with questions.
The clock, and what it takes away
Two clocks run at once and they are easy to confuse.
The first is contractual: policies require notification within a period and may set deadlines for submitting details. That clock is in your policy, it varies, and it is worth reading on day one rather than week six. It is also the one thing here that a broker will tell you in a minute.
The second is physical, and nobody sets it. Thermal receipts fade. Suppliers change systems and stop being able to send copies. People who remember what a cost was for leave. A cloud service you cancelled takes its history with it.
The second clock is the one that quietly decides how good your claim can be, because it is running before you know you will need anything. By the time a claim exists, part of the evidence has already degraded.
What follows from that is unglamorous: in the first week, gather and photograph everything, whether or not you yet know it matters. Sorting is cheap later; recovery is not possible later.
Extract a claim document
Upload an invoice, receipt or statement — supplier, dates, net, VAT, total and every line come back as rows, with the sum of the lines checked against the document total.
Upload a document
Drop your documents here
or browse to upload
Free: 10 pages / month · Upgrade for more
Files encrypted in transit · auto-deleted after processing · GDPR compliant
How it works
1 · Upload the file
Invoices, receipts, statements — up to 100 at a time. Photographs go through OCR first.
2 · Read by meaning
No template per supplier: a document you have never seen before works the first time.
3 · Check the arithmetic
Lines against totals, VAT against rate. A misread figure in a claim is worse than a missing one.
4 · Keep VAT separate
Net, VAT and gross as separate columns, so whichever basis applies is available.
5 · Review what is flagged
Only the uncertain fields — a faded receipt, a figure over a stamp. Not all hundred documents.
6 · Export the schedule
Excel or CSV, with the source file and page on every row.
Step 4 exists because the gross-or-net question comes up on almost every commercial claim and the answer depends on your VAT position and your policy. Keeping the three columns apart means you are not redoing the schedule when someone tells you which one applies.
Documents by claim type
| Claim | Documents that carry the weight | Usual weak point |
|---|---|---|
| Property damage | Purchase invoices, repair quotes and invoices | Nothing for older items |
| Stock loss | Purchase records, stock counts | Count dates that fit nothing |
| Business interruption | Accounts, bank statements, order book | No comparable period |
| Additional costs | Invoices after the event | Not separable from normal spend |
| Equipment breakdown | Service records, replacement quotes | Maintenance history missing |
| Goods in transit | Delivery notes, invoices, carrier records | Delivery note never kept |
Read the right-hand column as a list of things to fix before you need them. Every entry is cheap to arrange in an ordinary month and impossible to arrange in the week after a loss.
The business interruption row has its own guide, because proving what would have happened is a different exercise from proving what did — see how to evidence a business interruption claim.
The format problem, which is most of the problem
Claim documents do not arrive in a format. They arrive in seven, and the mismatch between them is what turns a day of work into a fortnight.
| Where it comes from | What it looks like | Why it resists a spreadsheet |
|---|---|---|
| Suppliers | PDF invoices, every layout different | No two put the total in the same place |
| The phone in someone's pocket | Photographs of receipts | Angled, shadowed, occasionally out of focus |
| The bank | Statement PDFs, one per month | Twelve files, one table needed |
| Portals | Downloads with generated filenames | Nothing in the name says what it is |
| Paper in a box | Scans made in one long batch | Several documents per file, in any order |
| Attachments buried in threads | Found by searching, and only if you remember | |
| Payroll and rotas | Exports in whatever the system offers | Different date conventions from everything else |
The fifth row is the one that catches people out mid-way through. A single scanned file containing four receipts has to become four rows, and the boundary between them is in the content rather than in the file. Splitting those by hand, halfway through a long afternoon, is where transcription errors are born.
What resolves all seven is the same operation: read each document for what it says rather than where things sit on the page, and give every resulting row a pointer back to the file and page it came from. After that the source format stops mattering, which is the whole point.
The schedule, and what makes one usable
Whatever the claim, what eventually gets handed over is a schedule: a list of items with amounts, and documents behind them.
Four things make one easy to accept, and all four are mechanical.
Every line ties to a document. Not a folder, a document — file and page. This single property does more than anything else to shorten the exchange.
The totals add up. Obvious, and routinely wrong when a schedule is assembled by hand from dozens of documents under time pressure.
Net, VAT and gross are separate. Then the basis can be changed without rebuilding anything.
Anything unusual carries a sentence. A part payment, a credit note, an item bought for two purposes. Explained by you, it is context; found by someone else, it is a question.
When the adjuster asks about one line
They will, and it is routine rather than adversarial. How it goes depends entirely on whether the working table still exists.
With it, a query is a filter and a reply the same day. Without it — if what was submitted was a total and the underlying rows were never kept — the same query means reconstructing how a figure was built, weeks after everyone moved on.
Three habits make the exchange short. Answer only what was asked; sending a whole file when three lines were queried creates work and widens the conversation. Disclose gaps yourself — a receipt you say is missing is handled very differently from one that is discovered. And reply completely rather than fast: a partial answer generates three more questions.
What an adjuster is actually looking for, and why some of these habits matter more than they appear, is on what loss adjusters look for in your records.
Duplicates and gaps, the two errors that survive review
Read a hundred documents carefully and you will catch a mistyped figure. You will not catch the document that is in there twice, and you certainly will not catch the one that is not in there at all. Neither error is visible from the document; both are visible from the list.
The duplicate. It usually arrives as a supplier statement or a chaser carrying the same amount as an invoice already captured, or as the same receipt photographed twice by two people. It is a legitimate document, correct in every particular, and counting it twice inflates the claim. Sorting by counterparty and amount finds these in seconds.
The gap. Harder, because absence leaves no trace. The only reliable method is to compare the document list against something that is definitionally complete — and for spending, that is the bank statement. Every payment that has no document behind it is a question, and most of those questions have an answer sitting in a portal nobody has opened.
This is why bank statements are worth reading into the same table as the invoices even though they prove nothing on their own. They are not evidence of the loss; they are the index that tells you what evidence should exist.
A third, quieter problem sits alongside these two: the same supplier appearing under three different names because someone typed the trading name once, the legal name once and an abbreviation once. Grouping by VAT registration rather than by name makes it disappear, which matters when the schedule has to be summarised by counterparty.
The first week, which decides most of it
In the week after a loss nobody has the appetite for administration, and that is precisely the week in which the quality of the eventual claim is largely determined.
Photograph everything, including things you think are irrelevant. Damaged items, the site, receipts, labels, serial plates. Storage is free and none of it can be recreated. Deciding what matters is a job for later.
Start a single folder, not a filing system. One place where everything lands, with no naming convention and no categories. Any structure imposed now will be abandoned within days and will cost you documents.
Write down what happened, dated, while it is fresh. Not for anyone else — for yourself in three months, when the sequence of events has blurred and you are being asked about it precisely.
Keep costs incurred because of the loss separate from the moment you start incurring them. A separate note, a separate folder, anything. This is the distinction that is nearly free now and expensive to reconstruct later.
Ask your broker what your policy requires and by when. One conversation, on day one. Notification periods are contractual and they are not the kind of thing to discover in week six.
Who in a business should actually do this
It defaults to whoever is least busy, which is usually the wrong answer, and in a small business it defaults to the owner — who is also dealing with the loss itself.
The work splits cleanly in two, and they suit different people. Gathering needs someone who was there and knows what things were: what was in that room, what that supplier had delivered, which costs are unusual. That knowledge cannot be delegated.
Assembling — turning documents into a schedule that adds up — needs someone methodical and can be delegated entirely, to a bookkeeper, an accountant or an assessor.
Recognising the split is worth something practical: the person who was there should spend their limited time on the part only they can do, and not on formatting a spreadsheet. The reverse arrangement is common and wastes the scarcer resource.
For larger losses, a loss assessor acting for you does the second part professionally and negotiates as well. Whether that is worth the fee is a commercial judgement, and it is a real option that many policyholders do not know exists.
Six mistakes
Sorting before gathering
In the first week, collect and photograph everything. Deciding what matters is cheap later; recovering a faded receipt is not possible later.
Submitting a total instead of rows
The total is accepted or queried. If it is queried and the rows were never kept, the reconstruction is the expensive part.
Filing payments apart from documents
The pair proves the item. Split across two systems, every line takes twice as long to present.
Mixing loss costs with ordinary spending
Nothing on an invoice says which it is. That distinction is made at the time or reconstructed at much greater cost.
Rebuilding the schedule when the VAT basis changes
Keep net, VAT and gross separate from the start and the question costs nothing.
Treating a query as an accusation
It is usually routine. A complete, fast, narrow answer closes it; a defensive or partial one extends it.