Expense fraud is usually small and repeated. The tests that find it are simple, if someone runs them.
Test expense claims for duplicates, missing or altered receipts, spend outside policy, split transactions under approval limits, weekend and holiday spend, round amounts, mileage that doesn't fit the journey, and items claimed that were also paid on a corporate card. Combine data tests on every claim with a risk-based sample and a small random sample, and record each finding.
Export all claims and card transactions for the period, with employee, date, merchant, category, amount, approver and receipt status. Run these tests across everything:
Data tests find patterns; receipts confirm them. Review receipts for:
Every card transaction should be matched to a receipt and a business purpose within a set number of days. Unreconciled transactions that stay open, and personal spend recovered late or not at all, are common findings.
For each finding record the claim, the issue, the amount, the outcome (repaid, corrected, accepted with reason) and any policy point it raises. Share themes with approvers, because most expense issues are approval issues. Repeat the data tests every month or quarter so the checks become routine.
Many of the same red flags apply to supplier invoices. See invoice fraud red flags.
UK: HMRC publishes approved mileage allowance rates; mileage paid above them can create a tax charge. US: the IRS sets a standard mileage rate each year, used by many employers as their reimbursement rate. Check your policy against the current published rate.
Score a payment request against the warning signs of fraud. Free to use online.
The Expense Claims & Corporate Card Audit System tests for exact duplicates, duplicates across employees and split purchases, flags approval problems, missing receipts, late submissions and blocked merchants, and matches card claims to the card statement, with an Expense Audit Playbook (PDF). Excel, no macros, one-time price.
Test every claim with data, then review receipts for all flagged claims plus a risk-based and a small random sample. Size the random sample to your volume; it's there so every claim could be checked.
One purchase broken into two or more smaller claims or card payments to stay under an approval or receipt limit.
Run the data tests monthly or quarterly. A short, regular review deters problems far better than a large annual one.
Last updated 18 October 2026. General guidance for finance teams; check it against your own policies, systems and local rules.