What is actually being proved
Everyone understands the first half instinctively: something happened, trade stopped, money was lost. The records for that half usually exist somewhere, even if they are scattered.
The second half is the one that catches businesses out. A business interruption claim asks you to establish what the business would have done in the same weeks had nothing gone wrong — and then to show that the difference between that and reality is attributable to the incident rather than to a slow quarter, a lost contract or a seasonal dip that happens every year.
That is a comparison, and a comparison needs two sides. Most of the work in evidencing a claim is building the side that never happened, from records of periods that did.
This has a practical consequence worth stating up front: a large part of the evidence for a claim is historical. It is not created during the disruption, it is retrieved from before it. Businesses that keep tidy monthly records find this stage takes a day; businesses that do not find it takes three weeks, and it is the same three weeks in which they are also trying to trade again.
What this guide is not
This is a guide to organising records. It is not financial, legal or insurance advice, and it deliberately avoids telling you what your policy covers — because that depends on wording that varies between policies and is often the single most consequential thing in the entire claim.
We do not assess or value claims
What a loss is worth under your policy is an assessment. It belongs to your broker, a loss assessor acting for you, or your solicitor.
We do not interpret policy wording
Indemnity periods, sub-limits, exclusions and the basis of settlement are legal questions with real money attached to the answer.
We do not submit anything to an insurer
The schedules you build are yours. Presenting them, and standing behind them, is done by you or by whoever represents you.
We do not verify that documents are genuine
Reading a document tells you what it says, not whether what it says is true. That distinction matters more in claims than almost anywhere else.
What is left after those four exclusions is still the majority of the work, and it is the part nobody else can do for you: finding the documents, reading them into rows, and arranging those rows so they can be checked.
Three decisions to make before you start
Each of these takes ten minutes now and saves days later, because each one determines the shape of everything you build afterwards.
One: who owns the file. A claim assembled by four people in four formats takes longer to reconcile than it took to gather. One person holds the master schedule; everyone else feeds it.
Two: what the unit of the schedule is. One row per document, always. Not one row per supplier, not one row per week. Rows can be summed into any grouping later; a summary cannot be split back into documents.
Three: where the originals live. One folder, one naming convention, and a column in the schedule pointing at the file. This sounds trivial until an adjuster asks for the document behind row 214 and you are opening PDFs one at a time to find it.
None of these decisions are about insurance. They are about making a body of evidence navigable, and they are the difference between answering a query in two minutes and answering it in two days.
1 · Fix the dates before you gather anything
The indemnity period is defined by the policy, not by how long the disruption felt. Find it, write it down, and arrange every schedule around it — because a schedule built on the wrong window has to be rebuilt entirely, and rebuilding is the most demoralising work there is.
In practice there are usually four dates that matter, and they are rarely the same: when the incident occurred, when trading was actually affected, when the indemnity period starts under the policy, and when it ends. Businesses tend to record the first and reason from it. Claims are settled on the third and fourth.
Write all four at the top of your master file, along with where each one comes from. When a figure later looks wrong by a month — and one will — this is the note that tells you whether the figure is wrong or the window is.
One more practical point on dates. Documents carry several of their own: the invoice date, the date the service was delivered, the date it was paid. For a claim, delivery is usually what matters, and it is the one least often recorded. Capturing all three from the start costs nothing; deriving one of them later means reopening every document.
2 · Build the baseline from real records
The baseline is the counterfactual: what the business would have done in the indemnity period without the incident. It is the part of a claim most often built on assertion, and the part most reliably tested.
Use at least two prior years, three if they exist. One year is not a baseline — it is a data point, and it cannot tell anyone whether a movement is a trend, a seasonal pattern or a fluke. With two years you can see direction; with three you can see whether the direction is real.
| Source | What it establishes | Weakness on its own |
|---|---|---|
| Filed accounts | The figures already stated publicly | Annual, so no monthly shape |
| Monthly management accounts | Seasonality and trend | Internal, so easier to question |
| Sales invoices raised | What was actually sold, and to whom | Says nothing about what was collected |
| Bank statements | What was actually received | Timing differs from when it was earned |
| Till or POS summaries | Daily trading pattern | Rarely retained far enough back |
| Order book or bookings | Demand that existed before the loss | Not all of it would have converted |
The right-hand column is why a strong baseline uses more than one source. Filed accounts are hard to argue with and give no monthly shape; management accounts give shape and are internal. Together they do the job neither does alone: the shape comes from one, the credibility from agreeing with the other.
Adjust the baseline for anything you know about that has nothing to do with the incident — a contract won, a price rise, a site opened, a customer lost. Doing this yourself, and saying you have done it, is far stronger than having someone else point it out. A baseline that has obviously been adjusted downwards where honesty required it earns credibility for every figure that follows.
3 · Separate the loss from ordinary variation
Every business has months that are down for no dramatic reason. The question a claim has to answer is which part of the shortfall is the incident and which part is the ordinary noise of trading.
The cleanest way to show this is to put the affected period next to the same months in prior years, and next to the months immediately before the incident. If the drop is visible against both, it is the incident. If it is only visible against one, expect the question.
It is also worth isolating anything else that moved in the same window. A supplier price increase, a marketing campaign that ended, a large customer that changed terms: these affect the numbers and have nothing to do with the loss, and pretending they do not exist is the fastest way to make an entire schedule look unreliable.
Where part of the shortfall genuinely cannot be attributed, say so and quantify it. A claim that concedes fifteen per cent to ordinary variation is far more persuasive than one that claims a hundred per cent of a movement nobody believes.
4 · Gather the revenue evidence
Revenue is where a claim is won or lost, and the strongest position is not one perfect source — it is two independent sources that agree.
Invoices raised tell you what was sold. Bank receipts tell you what was collected. They will never match exactly, because of timing and credit terms, but they should tell the same story about the shape of trading. When they do, both become more credible. When they diverge without explanation, both become questionable.
For retail and hospitality the equivalent pair is till summaries and banked takings. For subscription businesses it is the billing system and the receipts. In every case the principle is identical: one source is an assertion, two sources that agree are evidence.
Practical warning about formats. Revenue evidence usually arrives as PDFs — statements from the bank, exports from a till system that no longer exists, invoice copies from an old accounting package. Getting these into rows is mechanical work that expands to fill the time available, which is why it is worth doing in one pass rather than a document at a time as questions arrive.
5 · Gather the cost evidence, including what stopped
Costs split into three groups during a disruption, and a schedule that shows all three reads as complete. A schedule that shows only the first reads as optimistic.
Costs that continued regardless — rent, insurance, core salaries, finance payments. These are the ones a loss of profits calculation usually turns on.
Costs that stopped or fell — raw materials, variable staff hours, delivery, utilities at a closed site. Normally deducted, and always better volunteered than discovered.
Costs that only exist because of the incident — the third group, large enough to deserve its own step below.
The middle group is the one businesses skip, and skipping it is a mistake of strategy rather than honesty. Savings will be identified by somebody. Identifying them yourself costs the same money and buys credibility that applies to every other number in the file.
Keep net, VAT and gross separate on every cost row. Which basis a claim is settled on depends on your VAT position and your policy, and a schedule that holds all three can answer either question without being rebuilt. A schedule that holds only gross cannot.
6 · Capture increased costs of working as they happen
This is money spent specifically to reduce the loss: temporary premises, hired equipment, expedited freight, overtime, an agency covering an unavailable team. Most policies allow it within limits, and it is the most systematically under-claimed category in business interruption.
The reason is structural rather than careless. These costs arrive in small pieces, paid in a hurry by whoever was closest to the problem, often on a personal card, frequently without an invoice being requested. Nobody is thinking about evidence while they are keeping the business alive — which is exactly why this step has to be a habit rather than a task for later.
| Spend | Evidence to keep | What is usually missing |
|---|---|---|
| Temporary premises | Licence or lease, invoices, payments | The comparison to what it replaced |
| Equipment hire | Hire agreement and invoices | Why hire rather than repair |
| Expedited delivery | Carrier invoices showing the premium | The normal rate to compare against |
| Overtime and agency staff | Payroll records, agency invoices | Which hours are incident-related |
| Subcontracted production | Purchase orders and invoices | Evidence in-house capacity was lost |
| Emergency purchases | Card receipts and statements | The receipt itself, most of the time |
Notice the pattern in the right-hand column: what is missing is almost never the spending itself, which shows on the bank statement. It is the reason. An increased cost of working has to be shown to be an alternative to a larger loss, and that argument needs the comparison alongside the receipt.
A single sentence written at the time carries this. Recorded the same week — what it replaced, what it avoided, who authorised it — it costs thirty seconds. Reconstructed a year later from a card statement, it is a memory exercise for someone who has moved on.
7 · Build the schedule that everything else refers to
Everything above produces documents. This step turns them into a single table, and it is the artefact the rest of the claim is built on.
| Column | Why it is there |
|---|---|
| Document date and delivery date | A claim usually turns on when the service happened, not when it was billed |
| Counterparty and their VAT number | Names are written three ways; registration numbers are written one way |
| Document number | The only reliable way to spot the same document twice |
| Net, VAT and gross | Which basis applies is decided later, by someone else |
| Category | Continuing cost, saved cost, increased cost of working, or revenue |
| Attribution note | One sentence on why this row belongs in the claim |
| Source file and page | Turns a row back into a document in two seconds |
The last two columns are the ones that get left out and the ones that pay for themselves. A claim generates questions of exactly one form: “what is this, and where did it come from”. Those two columns answer it without anyone opening a folder.
Run the arithmetic checks before anyone else does. Line items against document totals, VAT against the applicable rate, schedule totals against the management accounts they should reconcile to. A discrepancy you found and explained is a footnote; the same discrepancy found by an adjuster is a reason to check everything else you submitted.
8 · Answer the questions before they are asked
By this point you know where your own file is weak. Every claim has soft spots: a month with poor records, a figure that depends on one source, an assumption in the baseline that could reasonably be challenged.
Address them in the submission. A short note saying “March till records were lost with the system; the figure is reconstructed from banked takings and the method is set out below” converts a discovery into a disclosure. The number does not change. What changes is whether the person reading it wonders what else has not been mentioned.
This is also the cheapest speed improvement available to you. Claims are rarely delayed by the size of the first submission; they are delayed by rounds of questions, each of which costs a fortnight of waiting on both sides. Pre-empting four questions can remove a month from the timetable.
What the recipient tends to look at first is set out in what loss adjusters look for in your records, which is worth reading before you finalise the file rather than after.
A worked example
A café with two sites loses the larger one for eleven weeks after a flood. Turnover across the business had been growing steadily; the smaller site stayed open throughout and took some of the displaced trade.
Dates. The flood is on 4 March. Trade stops the same day, the policy indemnity period runs twelve months from the date of loss, and the site reopens on 20 May. All four dates go at the top of the file, with a note that the claim period is not the same as the closure period.
Baseline. Three prior years of monthly takings, from till summaries cross-checked against banked receipts. March to May shows a consistent seasonal rise. The trend across the three years is upward, so the baseline is uplifted — and the uplift is stated as an assumption rather than buried in a total.
Separating the loss.The second site’s takings rose during the closure. That increase is deducted, because it is displaced trade rather than new trade. Nobody asked for this deduction; making it unprompted is what makes the rest of the schedule credible.
Costs. Rent, insurance and salaried staff continued. Food, casual hours and utilities at the closed site fell, and the reduction is quantified from supplier invoices and payroll. Both directions appear in the same schedule.
Increased costs. Equipment hire for the reopening, a fortnight of extra hours at the smaller site to handle displaced customers, and expedited delivery of replacement stock. Each row carries a one-line note explaining what it avoided.
The gap. Two weeks of till data is unrecoverable because the terminal was destroyed. Rather than estimating quietly, the file says so and reconstructs the fortnight from banked takings and card settlement reports, with the method written out in three sentences.
The finished file is a schedule with roughly four hundred rows, each pointing at a source document, plus a two-page note explaining the baseline, the deductions made voluntarily and the one gap. The arithmetic can be checked by anyone in an afternoon — which is the entire objective.
When two sources disagree
They will. Invoices raised and cash received never match exactly, till takings and banked amounts differ by float and timing, and the management accounts rarely agree with either to the penny.
The mistake is to pick the source that gives the better answer and quietly drop the other. Anyone experienced will ask about the source you did not use, and having dropped it is worse than the difference ever was.
What works is to show both, state the difference, and explain it in a sentence. Timing differences, credit terms, a deposit banked late, a refund netted off — these are ordinary and explainable. An unexplained difference is a problem; an explained one is just accounting.
Where the difference cannot be explained, size it. “Sales invoices exceed receipts by 4,120 across the period, of which 3,700 is a single customer paying in the following month; the remaining 420 is not identified” is a complete answer. Leaving the 420 to be found by somebody else is not.
What decays if you wait, and what does not
Not all evidence ages at the same rate, and knowing which parts are perishable tells you what to do first when there is more work than time.
| Evidence | Shelf life | Do this now |
|---|---|---|
| Photographs of the damage | Gone once the site is cleared | Photograph everything, dated, before clean-up |
| Thermal receipts | Illegible within months | Scan or photograph the day they arrive |
| Supplier duplicate invoices | Usually available for a year or two | Request in the first weeks, not the last |
| Reasons and authorisations | Fade with staff turnover | One sentence per unusual spend, at the time |
| Till and POS exports | Lost if the system is replaced | Export before any system is decommissioned |
| Bank statements | Effectively permanent | Can safely wait |
| Filed accounts | Permanent | Can safely wait |
The bottom two rows are the reason this table is worth having. Bank statements and filed accounts are the sources people rush to first because they are familiar — and they are precisely the ones that will still be there in a year. The perishable evidence is at the top, and it is the evidence nobody can reissue.
If you only act on one row, make it the fourth. Written reasons are free at the time, impossible later, and they are what turns a payment into an increased cost of working rather than an ordinary expense.
Queries you can predict
Almost every query on a business interruption claim is one of six. Reading them as a list before you submit is a cheap way to find the gaps in your own file.
Why this baseline? — because it is where a small assumption becomes a large number, and it is the first thing anyone tests.
What would have happened anyway? — the ordinary variation question, better answered by you than conceded later.
What did you save? — costs that stopped, which are always looked for and rarely volunteered.
Why was this cost necessary? — the increased-cost-of-working test, which needs the reason and not just the receipt.
Where is the document behind this figure? — the reason every row needs a file reference.
Why does this not agree with your accounts? — timing, accruals and VAT basis, all explainable in one sentence each.
None of these are hostile questions. They are the questions anyone would ask of a set of numbers they did not produce, and a file that answers them in advance is simply a file that has been finished.
Common mistakes
Starting with the total. A number decided first and evidenced backwards produces schedules that almost reconcile, which is worse than schedules that plainly do not. Build from documents upward and accept the total you get.
Claiming a hundred per cent of every movement. Some of the shortfall is ordinary trading, and a claim that concedes none of it invites scrutiny of the parts that are genuinely attributable.
Leaving out the savings. They will be found. Volunteering them costs the same money and buys credibility across the whole file.
Summarising too early. A weekly summary cannot be broken back into documents. Keep the rows and summarise on top of them.
Waiting until the disruption ends. The perishable evidence is being lost while you wait, and it is the evidence that cannot be recreated at any price.
Rounding. A schedule full of clean hundreds reads as estimated even when it is not. Exact figures from documents look like what they are.
Letting four people keep four spreadsheets. Reconciling them takes longer than gathering the documents did, and the reconciliation itself becomes a source of error.
Checklist
Four dates recorded, with the source of each: incident, trading affected, indemnity period start, indemnity period end.
Two to three prior years of monthly figures, from at least two independent sources.
Baseline adjustments listed explicitly, including the ones that reduce the claim.
Revenue evidence from two sources, with the difference between them explained.
Continuing costs, saved costs and increased costs of working in separate categories.
Net, VAT and gross held as separate columns on every row.
A one-line reason attached to every increased cost of working.
Source file and page on every row of the schedule.
Arithmetic checks run: lines to totals, VAT to rate, schedule to management accounts.
Known gaps disclosed in writing, with the reconstruction method stated.
Perishable evidence captured: photographs, thermal receipts, system exports.
One owner for the master file, one folder for the originals.
When to bring in help
Some claims are a fortnight of tidy work by one person. Others are beyond what an owner-manager should be doing while also rebuilding a business, and knowing which one you have is worth an hour of thought early.
Signals that point towards professional help: the figure is large relative to the business, the policy wording is genuinely ambiguous, the indemnity period is long, several sites or entities are involved, or the first response you receive is a detailed challenge rather than a request for documents.
A loss assessor acts for you, as distinct from the adjuster, who does not. Your accountant is the right person to make the schedules reconcile to filed accounts. A broker who placed the policy usually knows the wording better than anyone else you have easy access to.
None of them removes the work described here. What they do is decide what it means — and they all work faster, and cost less, when the documents already exist as rows with references attached.