Financial anomaly

An anomaly is not an accusation. It is an entry that falls outside the expected pattern and has no ready explanation attached to it, and what makes one visible is comparison. Inside a single ledger most anomalies look perfectly normal, because the ledger is internally consistent by construction: every entry balances and nothing openly contradicts anything else. Put the ledger, the bank statement and the e-invoice data side by side in one table and whatever is present in one source and absent from another stands out on its own. Turkey helps here, because electronic invoicing is mandatory for companies above the turnover threshold set by the tax authority, so for a large share of business to business transactions a third record exists outside your own books and outside the bank. Where two sources agree and the third is silent, the usual explanation is that nobody ever opened the third.

In practice the flags fall into a few families. Missing documents: an invoice that appears on the counterparty statement and nowhere in your ledger, or the reverse. Duplicates: the same invoice entered under two numbers, or the same payment loaded once by hand and once by an automated feed. Out of pattern amounts: an item far larger than anything else on that account, an unusual cluster of round numbers, or entries bunched into the last days of a period. Unexplained differences: the residual gap between two balances that cannot be tied to any known cause such as an advance, a credit note, an exchange difference, a value date shift, or the Turkish VAT withholding rule under which the buyer pays part of the tax directly to the tax office. Timing anomalies, where one transaction lands in different periods on the two sides, form a family of their own.

Two published figures frame the size of the problem. APQC puts duplicate and erroneous payments at 0.8 percent of annual disbursements for top quartile organisations and 2 percent for the bottom quartile, which is why a systematic scan tends to pay for itself. On the fraud side, the 2024 ACFE report records account reconciliation as the way 5 percent of cases were first spotted, with tips far ahead at 43 percent; those figures come from the 2024 edition, and the fourteenth edition was published in 2026. Read them carefully, because the distinction that matters most is that the method which first surfaces something is not the method that resolves it. Reconciliation produces a signal; people carry the investigation and the decision, and it cannot be said to prevent or resolve fraud.

Two further misreadings are common. The first is treating every flag as bad faith: most flagged items turn out to be data entry errors, late postings or format problems, and they close as soon as somebody looks. The second is setting thresholds so tight that the list becomes unreadable. A useful anomaly list is short, reasoned and closeable, with the reason for the flag and the name behind each decision recorded next to the line.

For a group finance function the practical question is which flags travel upward. A local team can absorb a keying error without telling anyone, and should. What belongs in the reporting pack is the narrow category that changes a number the group relies on: a duplicate payment already out of the bank, a supplier invoice missing from the statutory books, a difference against a counterparty that has stayed open across two consecutive closes. Agreeing that shortlist in advance is what keeps anomaly review from becoming either a monthly list nobody reads or a set of surprises that surface first in the audit.