Keeping Books a Stranger Could Audit
Five numbers, 253 documents, one practice year. The test that matters, the one step an outsider controls, and why entries are written twice.
There is one test for a set of books, and it has nothing to do with software. Could somebody who has never met you follow the paper from a bank statement back to an individual transaction, without asking you a single question? If yes, the books work. If no, they are a private diary that happens to contain numbers. On the closed practice year below, five summary figures sit on 253 documents, and every figure traces to a specific dated piece of paper somebody could pick up and check.
The worked example: five numbers, 253 documents
| Line | Documents behind it | Amount |
|---|---|---|
| Money in, twelve months | 138 dated sale lines | +$10,056.00 |
| The slice the card took | 12 monthly settlement statements | −$221.84 |
| Materials bought | 19 supplier receipts | −$2,554.50 |
| Costs that happen anyway, plus paid hours | 72 receipts and signed records | −$2,760.00 |
| What the year put in the account | 12 monthly closes against 12 statements | +$4,519.66 |
Check it: $10,056.00 less $221.84, less $2,554.50, less $2,760.00 is $4,519.66. The fourth row splits into $1,440.00 of costs that arrive regardless and $1,320.00 of paid hands, across 24 table receipts, 12 site and reader charges, 24 for fuel, and 12 signed records of hours paid.
Notice the second row in particular. Those settlement statements were written by somebody who has never met the owner, which is exactly why they count for more than anything written in-house.
Every event has one home
Decide the account list once, keep it short, and then never file anything anywhere else. Around 6 named accounts is usually plenty for a venture this size:
- What comes in.
- What things cost to make.
- What runs anyway, whether or not anything sells.
- What is owned.
- What is owed.
- The owner's own money, in and out.
The discipline is not the number of accounts. It is that a transaction which belonged in one of them last March still belongs in that one this March. Books that get reorganised mid-year are books that cannot be compared to themselves.
Why every entry is written twice
The account went up and a sale happened. Canvas arrived and the account went down. Each event lands in two places and the two have to agree.
That is not bureaucracy, it is error detection. When you total every account at the end of a period, the two sides have to come out equal. They will not, the first few times, and that failure is the entire point: a set of books that cannot disagree with you is not checking anything.
Software does this for you, which is genuinely useful and also means the failure is easier to ignore. If your books never disagree with you, find out whether that is because they are right or because nothing is being checked.
The one step an outsider controls
Everything above was written by you. Exactly one step brings in evidence you did not create: tying your own cash figure to the statement somebody else issued.
On the practice year that is 12 monthly closes matched against 12 statements, line by line. Every month's close was tied to that month's statement, which is why the year needed no detective work at the end of it.
Any difference gets investigated, never quietly adjusted. Adjusting a difference to make the numbers agree destroys the only independent evidence in the whole system, and it is the single most common way a set of books stops being worth anything.
One sale, all the way through
- The document exists first. A receipt, an invoice or a settlement line, with a date on it.
- It becomes an entry with two sides. What went up and what went down, same amount, same date.
- Each side lands in its account. The one home it was always going to have, decided in advance.
- The accounts are totalled. Both sides have to agree exactly, or you go hunting.
- Cash is tied to the statement. The one figure a stranger issued, matched line by line.
Step one is the one that gets skipped, and skipping it is how a business ends up with entries nobody can substantiate. The document comes first, always. An entry with no document behind it is a claim.
Why the paper count matters more than the number count
Look again at the document column. 138 sale lines and 19 supplier receipts describe the same year. Materials arrive by the bolt and hardware in runs, so nineteen pieces of paper cover a year of buying, while every individual sale needs its own dated line.
Fewer receipts, each one bigger, each one still needed. That asymmetry is worth understanding because it tells you where the effort goes: capturing sales as they happen, and filing supplier paper as it arrives. Those are two different habits with two different failure modes.
What good books make possible
- An honest hourly rate. The top of that fraction is a year's revenue and costs, and it is only trustworthy if the books are. The division is here.
- A price you can defend. A costing card needs real supplier prices and real times, both of which live in the records.
- A conversation with a lender or a buyer. Both of them are strangers, which is the whole test.
- A return you can substantiate. Which is the same discipline households need, worked out in the proof column on a personal return.
The five failure modes, and what each one looks like
| Failure | How it shows up |
|---|---|
| Mixed personal and business money | the paper cannot separate, at any skill level |
| Entries with no document | a number nobody can substantiate |
| A cash figure adjusted to agree | the only independent evidence is destroyed |
| Accounts reorganised mid-year | the year cannot be compared to itself |
| No monthly close | a January spent reconstructing twelve months |
Every one of the five is a habit rather than a skill, which is the good news. None of them requires an accountant to prevent, and all of them are expensive to fix afterwards.
Where the discipline usually breaks
Not in the arithmetic. In three habits.
Capturing cash and small sales. Nobody else is holding a copy, so the record is entirely yours to create and entirely yours to lose. Dated on the day, not reconstructed later.
Filing supplier paper. It arrives at inconvenient moments and it is boring, and eighteen receipts filed with one missing is a year with a hole in it.
Doing the monthly close. The close is what stops a year becoming a detective exercise in January. On the practice year, 12 monthly closes are the reason the fifth row of the table above needed no reconstruction at all.
What monthly actually means
The monthly close is the habit that carries the whole system, and it is smaller than it sounds. On the practice year it is 12 closes across twelve months, each one matching that month's own figures against that month's statement.
In practice it is about an hour: file the paper that arrived, enter anything not yet entered, total the accounts, and tie the cash figure to the statement. An hour a month is twelve hours a year, against the alternative, which is a January spent reconstructing a year from memory and a shoebox.
It also means an error is at most a month old when it is found. A mistake found in the same month is a correction. The same mistake found eleven months later has been quietly wrong in every figure computed since.
What this page will not tell you
Which method your situation requires, what has to be filed and when, what records are required to be kept and for how long, or how any of it is treated for tax. Those depend on your situation, your entity, your state and rules that change, and they belong to a CPA and to irs.gov rather than to an article.
What an article can do is give you the test. Could a stranger follow the paper. If the answer is yes, every conversation with a professional starts from a much better place, and costs less.
The same is true at the front of the venture: a unit cost with real supplier prices behind it is a document, and costing one unit to the cent is where those prices first get written down.
One month. Whatever you cannot trace is the exact shape of the gap, and it is far better to find it in one month than in twelve.
Going further
Owning a Business counts the documents behind one closed practice year and shows what an outsider actually reads, then goes on to the questions books make answerable: paying yourself on purpose, what outside money costs, and what the thing is worth. Every figure is computed in code and no outcome is promised anywhere in it.
For what applies to your own venture, a CPA and irs.gov are the right sources, and sba.gov explains what free advisory help exists.
Questions people actually ask
What makes a set of books good?
One test: could somebody who has never met you follow the paper from a bank statement back to an individual transaction without asking you a question. On the practice year here, five summary numbers sit on 253 documents, and every one of them traces.
How many accounts does a small business need?
Fewer than people think. Around 6 named accounts is usually plenty: what comes in, what things cost, what runs anyway, what is owned, what is owed, and the owner's own money. The discipline is that nothing may be filed anywhere else afterwards.
Why are entries written twice?
So an error announces itself instead of hiding. Each event lands in two places and the two have to agree, and when you total every account at the end of a period the two sides have to come out equal. A set of books that cannot disagree with you is not checking anything.
What is the most important single step?
Tying your own cash figure to the statement somebody else issued. Everything before it was written by you. Any difference gets investigated rather than quietly adjusted, because adjusting it destroys the only independent evidence you had.
Is this accounting advice?
No. This is general financial education using an invented practice venture. It is not financial advice, not accounting advice, and not a recommendation about how to keep your own records. Requirements vary, and a CPA and irs.gov are the right sources for what applies to you.