delbar · fraud signals
Where a company's payments deserve a human look
Delbar runs a payment journal through ten checks that internal auditors use and shows where to start. The data stays in this window and is sent nowhere.
what we check
Which columns the file needs
Payment journal, required: date, amount, payee. Recommended: time, document, tax_id, account, purpose, type (supplier / salary), created_by, approved_by, registered, phone.
Employees: name, account, phone, terminated. Separator: comma, semicolon or tab. Dates as 2025-03-18, 18.03.2025 or 03/18/2025. From Excel: “Save as → CSV UTF-8”.
Amounts from the limit up need the CFO's signature. Splitting and “just under the limit” are measured against it.
No results yet. Press “Find what looks suspicious” and Delbar checks a demo company whose journal has six fraud schemes planted in it. Then load your own journal and compare.
Signals by risk, highest first
| Risk | Signal | Who | Payments | Amount |
|---|
Pick a signal on the left to see why it fired and what to check.
benford's law
First digits of supplier payment amounts
In natural money data an amount starts with 1 far more often (30%) than with 9 (5%). When amounts are shaped by hand, for example kept just under the approval limit, the distribution breaks. This is a signal about the whole journal, not about a single payment.
A signal is a reason to look at the source documents, not proof of guilt. Most signals have a lawful explanation; a person decides after checking.