Most internal fraud is not discovered by looking at the profit and loss account. It is discovered when someone outside the company claims something the books say was already settled: a supplier who says they have not been paid, or the tax authority sending an enforcement notice for a tax that is recorded as paid.
That has a practical consequence: well-kept books are not just a legal obligation. They are where fraud leaves its trace, because whoever diverts money has to balance the entry somehow, and that entry always has something that does not fit. All it takes is to inspect the records with a critical eye and one question in mind: where did the money actually go?
Two real examples of how this works.
Example 1: the VAT that was paid twice, but only once to the tax authority
What the fraud is. The VAT return falls due. The person who runs the accounts files the return but, instead of paying it, requests a deferral from the tax authority. The money the company thought it was paying in tax leaves the bank anyway, with “Tax Authority” as the reference, into an account the employee controls.
How it is covered up. The accounting entry is impeccable: debit VAT payable, credit bank. The supporting document filed is the PDF of the return, which exists and is authentic. Nobody asks for the payment reference number because the entry balances and the document is there.
How it is detected. By looking at the original bank statement, not the ledger entry: the payee of the debit is not the tax authority. And by logging into the tax authority’s online office with the company’s own digital certificate: the “outstanding debts” query shows a live deferral on a period the books treat as settled. Two five-minute checks the employee cannot manipulate.
How it is prevented. The company’s digital certificate is held by the director, not by the bookkeeper. Every month, someone other than the bookkeeper reviews outstanding debts on the tax authority’s portal and matches every tax payment to its payment reference and to the actual bank debit.
Example 2: the customer who paid, and the receipt that became an expense
What the fraud is. A customer pays an invoice by transfer or in cash. The employee who handles receipts keeps the money. The customer has paid and has a receip. The employee’s risk lies elsewhere: the invoice is still outstanding in the sales ledger, and sooner or later someone will chase the customer for payment. The customer will reply that they already paid, produce their receipt, and everything will come out. So the employee needs to make that balance disappear before the reminder goes out.
How it is covered up. The balance has to vanish without going through the bank, because the money never came in. The easiest way is a manual journal: credit the customer, debit an expense account (“other services”, “differences”, “sundry expenses”). The customer balance is zero, nobody chases them, the bank reconciliation balances, and the amount is absorbed into the profit and loss account as just another expense. Variants: issue a credit note for a discount or return that never happened, or write the invoice off as a bad debt.
How it is detected. With one question to the general ledger: which entries reduce customer balances without bank or cash as the other side? A credit to a customer against an expense account has no normal commercial explanation; a credit note with no physical return and no sales approval, none either; a bad-debt write-off for a customer who keeps buying, even less. And the definitive confirmation comes from outside: the statement of account sent to the customer, whose reply says “I paid that invoice on the 12th”.
How it is prevented. Credit notes, discounts and bad-debt write-offs are approved by someone outside the receivables function. No manual journal may touch a customer account with an expense account as the other side without a supporting document and a second signature. And statements of account go out to customers from the system, with replies to a mailbox not managed by the person who applies receipts.
If you use Xero, you already have the reports
No forensic software is needed. Three standard Xero reports cover both examples:
- Bank Reconciliation Summary: shows unreconciled items and those that have been pending for a while. Any bank outflow with no matching entry, or sitting “in transit” for more than 30 days, is the first question to ask.
- Account Transactions is the key report, because it shows for each movement not just the amount and date but the contra account. Filtered on the tax payable account, it lets you match each payment to the bank statement and to the tax authority’s outstanding-debts query. Filtered on the generic expense accounts, it brings to light the debits whose other side is not the bank or a supplier but a customer: that is exactly the entry in example 2. The question is not how much was spent, but against what.
- History & Notes: Xero records who created or modified each invoice, credit note and manual journal, and when. A credit note issued at month-end by the same user who applies receipts shows up there with a name and a time.
Add Aged Receivables to see which invoices disappear from the outstanding list with no receipt attached, and the Journal Report to review every manual journal posted in the month, and you have a half-hour monthly review.
The rule that changes everything
Both frauds above have one thing in common: a single person set up third parties, kept the books, made the payments and reconciled the bank. With that level of control, any diversion can be covered up. With two different people, the fraud needs an accomplice, and accomplices are expensive and unreliable.
Whoever keeps the books must not be the one who makes the payments, and the bank reconciliation must be reviewed by a third person. In a small company that third person can be the director or the external accountant, with read-only access to the bank. It costs nothing. It only requires looking at the records with the right question: not whether they balance, but where the money went.