Finance Index
What are the most common expense fraud schemes, and how do you detect them?
Reference guide to expense fraud patterns detection, including card controls, policy design, employee spend workflows, receipt capture, and reconciliation.
The classic schemes are a short list: duplicate submissions (the same receipt claimed twice, often once on card and once as reimbursement), inflated or altered amounts, fictitious expenses with fabricated receipts, mischaracterized personal spend dressed as business, and mileage padding. Detection combines automated screening for the mechanical patterns (duplicates, round numbers, just-under-threshold clustering) with risk-based human review of the judgment calls - and the most powerful deterrent is simply telling people the screening exists.
At a Glance
| Aspect | Short Answer | Why It Matters |
|---|---|---|
| The most common expense fraud | The classic schemes are a short list: duplicate submissions (the same receipt claimed twice, often once on card and once as reimbursement), inflated or altered amounts, fictitious expenses with fabricated receipts, mischaracterized personal spend dressed as business, and mileage padding. | Keeps spend tied to policy, ownership, and review. |
| What red flags should automated | The reliably suspicious signals: exact-duplicate amounts and dates across submissions, round numbers (genuine receipts rarely total to even dollars), amounts clustering just under approval or receipt thresholds, weekend and holiday charges in roles that don't work them, sequential or reused receipt images. | Keeps spend tied to policy, ownership, and review. |
| Card control | Duplicates hide in the gap between systems - an employee expenses a dinner as a reimbursement and the same dinner also posts on a corporate card, or two attendees both claim the same shared meal. | Keeps spend tied to policy, ownership, and review. |
| Risk check | AI-generated receipts are a real and growing problem precisely because they're cheap and convincing to the human eye. | Keeps vendor records and payment decisions reliable. |
| How much does expense | Occupational-fraud research consistently finds expense reimbursement among the more common fraud schemes, though typically smaller per-incident than billing or payroll fraud; the meaningful cost is usually the slow accumulation of small padding rather than one large hit. | Keeps spend tied to policy, ownership, and review. |
What red flags should automated checks catch?
The reliably suspicious signals: exact-duplicate amounts and dates across submissions, round numbers (genuine receipts rarely total to even dollars), amounts clustering just under approval or receipt thresholds, weekend and holiday charges in roles that don't work them, sequential or reused receipt images, and merchants inconsistent with the stated business purpose. None of these is proof on its own - each is a reason to look. The job of automation is to surface the cluster; the job of the reviewer is to judge the pattern.
How do you detect duplicate expense claims across card and reimbursement?
Duplicates hide in the gap between systems - an employee expenses a dinner as a reimbursement and the same dinner also posts on a corporate card, or two attendees both claim the same shared meal. Catch them by screening across both the card feed and the reimbursement queue on amount, date, and merchant, and by flagging same-event same-amount claims from different employees. The structural fix is to shrink the gap: when most spend is on company cards inside one system, there's far less surface for a card-vs-reimbursement double-dip.
We caught an employee fabricating receipts with editing tools or AI - how widespread is AI-generated receipt fraud and what defenses exist?
AI-generated receipts are a real and growing problem precisely because they're cheap and convincing to the human eye. Defenses: prefer e-receipts and merchant-direct feeds over uploaded images where possible, match every receipt against the actual card transaction (a fabricated receipt with no matching charge fails immediately), screen for image-metadata and template anomalies, and lean on the card transaction itself as the source of truth - a receipt that can't be tied to a real authorized charge is the tell.
How should we respond when expense fraud is confirmed?
Preserve the documentation, involve HR early, scope how far back the pattern goes, pursue restitution, and reserve law-enforcement referral for material or criminal cases on counsel's advice. Apply the response consistently regardless of the employee's seniority - selective enforcement is what turns a fraud finding into a liability.
Does telling employees that expenses are sampled and analyzed reduce fraud on its own?
Yes - visible auditing is one of the better-supported deterrents in the fraud literature. The perception that spend is reviewed changes behavior more cheaply than reviewing everything does; announce the screening, publicize (anonymized) that it works, and the deterrent compounds.
A manager has been approving a friend's padded expenses for months - how do approval-collusion schemes get caught?
Collusion defeats single-approver controls by design, so catch it with controls the approver can't see around: independent risk-based sampling outside the approval chain, anomaly screening on approver-employee pairs (one approver consistently passing one employee's flaggable spend), and rotation or second-review on high-trust relationships. Collusion is found by analytics that don't ask the colluding approver's permission.
Stampli perspective
Stampli's structural answer to expense fraud is to remove the gaps it hides in. Expense Cards with employee-initiated requests, finance-defined guardrails (limits by cardholder, merchant category, or vendor), and mobile receipt prompts mean spend originates on controlled company instruments with documentation captured at the source - shrinking the reimbursement surface where duplicate and fictitious claims live. Real-time transaction posting surfaces anomalies in days rather than at statement close, and Stampli AI supports anomaly and duplicate surfacing across card and invoice workflows, with humans reviewing before financial action.