This is the third of three notes explaining our tiers, and the only one that starts with a refusal. The other two explain what you get. This one has to begin with the situation in which you cannot buy it at all.
It is written the way I would want a service explained to me before spending €950 on it: what it does, what it refuses to do, what it does when the numbers do not work out, and the several situations where your money is better spent elsewhere.
The refusal comes first: no documents, no Reality Check
This tier only starts when you are holding the seller's own paperwork. The minimum set is:
- Filed accounts or tax declarations, two to three years. What the seller told the state he earned.
- Business bank statements, twelve to twenty-four months. What actually arrived in the account.
- The till or card-terminal export, if it is a business that takes cash.
- The lease, and the licences as they are currently held.
- A staff list with roles, start dates and salaries — without names, ID numbers or addresses. I do not need to know who they are to work out what they would cost you, and I would rather not hold personal data I have no use for.
If that set is not there, the order does not become a Reality Check. It is delivered instead as the €550 Screen, together with a written list — in English and in Spanish or Portuguese — of exactly what to ask the seller for, in what words. When the documents arrive, it upgrades for the €400 difference, within 30 days.
The reason for the refusal is simple, and it is the whole difference between this tier and a report that merely sounds expensive. Everything here is arithmetic performed on real pages. Take the pages away and the only thing left to build on is assumption. A report built on assumption reads exactly like a report built on documents — same headings, same confident tone, same number at the bottom — and you cannot tell them apart. So I do not produce the second one.
There is a free diagnostic in this that costs you nothing. How the seller behaves while you are asking — how fast, how completely, whether he offers to let you "come and watch the till for a week" instead of sending the statements — goes into the memo. A seller who stalls for three weeks on bank statements has already told you something the statements might not.
Three stories about the same money
Here is the idea the whole tier rests on, and it is not complicated.
There are three separate accounts of the money a small business made last year, and they were written by three different parties for three different reasons:
- What the seller says — in the advertisement, in the meeting, in the spreadsheet they prepared for you.
- What they filed with the state — the tax declarations, where declaring more means paying more.
- What reached the bank — the statements, which they did not write.
Anyone can produce one of these. Producing all three so that they agree with each other takes considerably more effort, because a tax declaration was written for a reader who is not you, and rewriting it to suit a buyer has its own consequences for the seller.
Be careful how far you take that, though, because I nearly overstated it myself. A tax declaration is something the seller submitted. It is not something the tax office checked and approved. It can be amended. And three records that agree with each other perfectly can still all be wrong, if whoever prepared them was careful and had time. Comparing them raises the effort required; it does not close the door. Nobody selling you a document review can close that door, and you should be suspicious of anyone who says otherwise.
So the work is not "check whether the seller is honest" — that is not a question documents can answer. It is narrower and much more answerable: put the three accounts side by side and find the places where they disagree.
Both directions of disagreement matter, and buyers usually only worry about one of them.
If the filed figure is lower than what the seller claims — "the real takings are higher, some of it doesn't go through the books" — then you are being asked to pay for earnings that officially do not exist. You cannot verify them, you cannot bank on them, and if you pay for them you have paid for a sentence. That is not a moral point. It is a pricing point.
If the filed figure is higher than what the bank received, something else is going on, and the memo asks what.
There is a third case that people forget: the seller genuinely does not know. Small businesses are often run out of one head and a notebook, the accountant files what they are given, and nobody has ever laid the three records side by side. A disagreement between them is a question to be answered, not an accusation, and the memo puts it that way.
Two pieces of arithmetic every month has to survive
Before anything is rebuilt or compared, the documents themselves have to hold together. Two checks do most of the work, and both are simple enough to describe in a sentence.
One: each month of bank statements must add up on its own. Money at the start, plus money in, minus money out, equals money at the end — and the closing balance of one month must be the opening balance of the next. If a month does not add up, or if the months do not join on to each other, that is not a rounding matter. It usually means one of two things: a second bank account you have not been shown, or a statement that has been altered. Either way the work stops there until it is explained.
Two: the card sales must arrive in the bank. Card payments rung up on the till have to show up on the bank statement a few days later as a payment from the card company, minus its fee. This is the most useful check available on a deal of this size, because the till and the bank are two separate records of the same event and the seller controls only one of them. If card takings appear on the till and never reach the account, that is one of the most serious things a report of this kind can find.
There is a third, quieter one. A missing month is a finding, not an administrative detail. If eleven months of statements arrive and one does not, that month is the interesting one until proven otherwise.
And a boring but genuinely dangerous one: the decimal point. In Spain and Portugal 1.000,50 means one thousand euros and fifty cents, and 1,000.50 means the same thing in English notation. Read one as the other in the wrong direction and you are out by a factor of a thousand. Every figure is checked for this, because it is exactly the sort of mistake that survives all the way to the bottom of a page and looks perfectly plausible on the way.
What the business actually put in the owner's pocket
The declared profit of a small business is almost never the amount the owner lived on, and this is normal rather than sinister. The owner may have paid themselves a salary of €900 a month while the business supported them entirely; there may be a van on the books that is really the family car; a roof was replaced once, five years ago, and it sits in last year's costs; a relative works there unpaid.
So the declared result is rebuilt into a figure that means something to a buyer: the money the business would actually put in the pocket of an owner who works in it full time, before that owner pays themselves a wage. (The trade name for it is seller's discretionary earnings. You can forget the term; the sentence above is the whole of it.) Between four and eight named adjustments are made, depending on how good the documents are, and every single one carries a reference to the document and the page it came from. Typical lines are the declared result itself, normalising the owner's salary to what the job would actually cost to fill, removing genuine one-offs, and correcting the rent to what the lease says rather than what is in the accounts.
The point of the page reference is that you can hand this to your own accountant and they can check every line against the paperwork without asking me a single question. If a number cannot be traced to a page, it does not go in.
How many years to get your money back — and why that is not a price
Once there is a defensible earnings figure, one piece of arithmetic follows: the asking price divided by those earnings, which gives the number of years of owner earnings the price represents.
Three things about it, and all three matter more than the number itself.
It is printed as a range, not a single figure. Low, middle and high, because the earnings figure is itself a reconstruction and a single number invites you to treat arithmetic as a fact. The sentence in the memo reads like this: at €N to €N a year of owner earnings, the asking price equals M to M years — before your own salary.
It is before a great many things. Before your salary, before tax, before financing, before replacing equipment, before the working capital you will need on day one, and before any allowance for the risk of owning the thing. It is a floor for thinking, not a result.
It is not a valuation, and I am not permitted to let it become one. I am not a valuer. The words valuation, worth, supportable price and negotiation anchor are banned from our memos, deliberately, because each of them converts a piece of division into an opinion about your deal. If you need a certified valuation — for a bank, for a court, for a partner — you need a licensed valuer, and that is a different profession from this one.
When a check cannot be completed
This is the part I would want to read first if I were buying, so it gets its own section.
If the bank does not tie, or the till takings do not reach the account, or a month is missing, or the name on the documents is not the name on the register — the calculation stops. The memo does not produce a softer version of the answer. It says, in plain terms, this calculation cannot be completed on the documents supplied, and here is precisely what is missing.
There is a related outcome in the public-record part of the memo. Sometimes the honest result of a check is the public record does not answer this question, and I am not going to pretend that it does. We print that as a named result of its own — it is labelled cannot clear, which sounds like banking language and is not: nothing is being cleared or paid, it simply means the question stayed open.
It is deliberately not the same as "nothing found". "Nothing found" is a conclusion. "Cannot clear" is an admission. Reporting the second as if it were the first is the easiest way for a screening service to be quietly useless while appearing thorough, and it is why we gave the admission a name instead of a footnote.
An unwelcome result is not, by itself, grounds for a refund. The refund attaches to work not done or to a mistake of fact — not to an answer you did not want.
What else is in the memo
- The whole €550 Screen layer, re-confirmed — the public-record findings, the competition section, the verdict, the evidence annex with the date each source was read.
- A document map — what arrived, what each item proves, what is still missing, and which missing item is worth chasing first.
- A stress test — three scenarios priced in euros: the rent is renewed at the going rate, the key person leaves, the licence has to be re-granted.
- A negotiation plan — three to five levers in the order I would raise them, plus a walk-away line. Ammunition for your side of the table; the decision stays yours.
- The verdict, re-issued on the fuller evidence. Documents move it in both directions, and the memo says which way and why.
Who does the arithmetic — a question worth asking of anyone in 2026
A spreadsheet does the calculation. Software and language models are used to collect candidate figures from documents — a first pass, nothing more. A language model is never the source of a number in our memos. I re-compute every figure by hand and sign it.
I am explicit about this because the rest of our work is heavily automated and I would rather you knew exactly where the automation stops. A model that reads a scanned Spanish bank statement will occasionally return a figure that is confident, well-formatted and wrong, and it will never tell you which one that was. Arithmetic on somebody's life savings is not the place for that.
Things we do not do, although they would look impressive
There is a standard shelf of techniques that get mentioned in this field, and we have deliberately left them off. Naming them costs us the appearance of sophistication and I think it is worth it:
- Statistical tests on the digits of the numbers (Benford's law and similar). They need far more data than a small business generates, and seasonal trade sets them off. On this size of business they produce false alarms with an air of authority.
- Machine-learning fraud detectors. Trained on data that does not resemble a Spanish neighbourhood bar.
- Discounted cash-flow models and betas. Finance-textbook machinery applied to a business with two years of hand-kept accounts. The output looks precise and is decided almost entirely by the assumptions you feed in.
- A percentage chance the deal goes wrong. There is no honest way to compute one, and a number that cannot be computed honestly is worse than a sentence, because you will remember the number.
All four look clever in a report and none of them would change your decision. Instead the memo describes the ways the deal can go wrong in words, and prices the three most likely ones in euros.
"One person, five days. How is that enough?"
A fair question, and the answer is that the size of the job is not what people picture.
A neighbourhood bar or a small shop produces two or three years of accounts, twelve to twenty-four bank statements and a till export. That is a stack you can put on one desk. It is not a company audit, and nothing here pretends to be one: an audit tests everything and issues an opinion under professional rules, and it costs several times this. What happens here is a fixed set of checks — the ones described above — run over a known quantity of paper.
The three to five working days are also not spent reading. Most of the time goes on the parts that wait: a public register that answers slowly, a municipal enquiry, and the pass where every figure gets re-checked by hand. If it were purely reading it would be a day.
What that means honestly is this. A fixed set of checks over a fixed pile of documents catches the ordinary failures — the missing month, the second account, the takings that never arrive, the profit that exists only in conversation. It is not a search of everything, and a problem that lives outside those checks will not be found by them. If you want the whole of a business turned over, you want a firm and a five-figure budget, and for a €60,000 bar that arithmetic does not work for anybody.
"I could read these myself, or give them to my accountant. Why pay you?"
Both are real alternatives and I would rather set them out properly than pretend otherwise.
If you have an accountant who does buyer-side work on small business purchases in Spain or Portugal, use them. There is a genuine advantage on their side that I do not have: a regulated professional carries professional standards, a body they answer to, and usually indemnity insurance. I do not carry that insurance, and my liability is capped at the fee you paid — that is written in the terms and I am not going to bury it here.
What is different here is the direction of the reading. An accountant is trained to work with the documents in front of them, and is very good at it. This check spends most of its effort on what is not in front of it: the month that never arrived, the second bank account implied by a balance that does not carry over, the card takings that stop appearing in March, the fact that the name on the accounts is not the name that holds the licence. Those are absences, and absences are not what a set of books is designed to reveal.
The second difference is that the numbers land next to the public record and the market — the register, the licence position, the competition around that address, what comparable businesses ask. An accountant is not usually asked to assemble that, and it is where about half of the findings come from.
And the third: no relationship with the seller, no commission on completion, and my fee is the same whether you buy or walk. There are competent people around every deal whose income depends on it closing. I am not one of them, and at some point in a negotiation that turns out to be the useful thing about me.
As for reading them yourself — you can, and for some of it you should. The bank check in particular is arithmetic anyone can do at a kitchen table: take each month, add the money in, subtract the money out, see whether you land on the closing balance, and check that it matches the next month's opening. If you will genuinely sit down and do that for eighteen months of statements, and chase the missing one, then you have done the most valuable single check in this note for nothing. What is being sold here is not a secret method. It is the discipline to run every check even when the deal looks fine, and to write down the ones that failed.
Four situations where you should not buy this
- You do not have the documents yet. Then this is not the tier — and you would be told so rather than sold it. Start at the Screen, use the request pack, upgrade for the difference when the papers arrive.
- You need a certified valuation. For a bank, a court, or a partner who wants a signed figure — that requires a licensed valuer. This produces arithmetic on documents, and it says on its face that it is not a valuation.
- You want proof that the documents are genuine. That cannot be delivered, by me or at this price. The three records are checked against each other, which makes careless faking easy to spot; it is not proof that any of them is real.
- You have already decided, and you want the contract drawn. That is a lawyer's work, and a good one is cheaper than the mistake this would have caught earlier.
The honest edges. Not an audit and not a valuation: a common-sense test on numbers nobody has audited. No visit to the premises, no forensic accounting, no legal opinion, no negotiating on your behalf, no support after completion, and no promise that the seller's documents are true. The documents are compared with one another; they are not checked with the banks, tax offices or councils that issued them. Carefully prepared false paperwork is named in every memo as a blind spot — comparing the three records makes it harder to get away with, and never impossible. There is no professional indemnity insurance behind this work and liability is capped at the fee you paid. One set of documents per report: your evidence is fixed at the point we confirm it, and anything that arrives afterwards goes into a re-issue or a fresh report rather than being quietly folded into the one already running. I never contact the seller or the broker — the documents travel from the seller, to you, to me. Commercial opinion and general information, not legal, tax or financial advice.
Where this fits. Quick Scan (€99) reads the advertisement against the market it came from, using our own algorithm, in minutes. Deal Screen (€550) is where a person reads the public record around a named business and signs the memo. Reality Check (€950) is the same, plus the seller's own documents tested against each other — and it only starts when you are holding them.
Check my deal →This note describes our own published tier specification as it stands on 3 August 2026. Figures quoted for prices, delivery times, document requirements and the upgrade credit are our commercial terms, not measurements. Legal and accounting mechanisms are described in general terms and their application depends on the facts of a transaction and on the region. General information and commercial opinion, not legal, tax or financial advice.