FRAUD CATEGORY 1 OF 6

Duplicate expense claims. Caught before approval, not after audit.

The same receipt submitted twice is the most common fraud pattern in expense management. Sometimes it is deliberate. Sometimes it is an honest mistake. Either way, once approved, the money is out and recovery takes weeks or never happens at all. REME catches duplicates at submission time, before approval.

Under 200ms per check Cross-employee detection Image forensics + pattern matching Confidence-scored

How duplicate expense claims actually happen

Duplicate claims are not always what you would expect. They come in patterns that manual review consistently misses. Three of the most common:

Same receipt, different angles

An employee photographs the same receipt from three different angles and submits each as a separate claim. The image is the same, the metadata is different. Human review catches maybe one in five. Our engine catches all three.

Split between employees

Employee A and Employee B both attended the same client dinner. Employee A submits the receipt. Employee B submits it too, claiming his portion separately. Both employees expect to be reimbursed. Human review at the manager level catches this almost never. Our engine catches it because we look across employees, not just within one.

Same receipt with small modifications

The receipt is edited slightly: amount changed by 10 rupees, date shifted by a day, item description reworded. Looks different to human review. Our engine matches on the underlying pattern (same vendor, same core amount, adjacent dates) and flags the similarity.

SEE IT IN ACTION

See duplicate detection in action

How the same receipt gets caught even when submitted from different angles, by different employees, or with small modifications.

How our engine actually detects duplicates

Duplicate detection runs four parallel checks on every claim submission. Any single check triggering a match flags the claim for finance review. All four:

01

Image forensics

Perceptual hashing of the receipt image. Detects the same photo even when cropped, rotated, or slightly re-lit. Not defeated by different angles.

02

Amount pattern matching

Same amount from the same vendor within a rolling window flags for review. Configurable window (default 30 days). Configurable tolerance for near-matches (default 5 percent).

03

Vendor + date proximity

Same vendor identity on adjacent dates flags for review. Common pattern for employees splitting one meal receipt across two days.

04

Cross-employee reconciliation

Same receipt image, amount, or vendor pattern submitted by different employees within a rolling window flags all involved claims for finance review.

What happens when we flag a possible duplicate

Finance sees the flag with the specific reason. Examples:

"Possible duplicate of receipt submitted 3 days ago by Employee A. Match confidence: 96 percent. Image hash matches. Vendor matches. Amount matches within 2 percent."

Finance can approve or reject in one click, optionally messaging the employee for clarification. Most flags resolve in minutes.

"Possible split duplicate. Same vendor, same date, similar amount claimed by Employee A and Employee B. Both claims flagged for review."

Finance can approve one and reject the other, or approve both if the split is legitimate. The transparency is intentional.

"Near-duplicate. Similar vendor and amount to receipt submitted 12 days ago. Match confidence: 78 percent. Recommended for manual review."

Lower confidence flags get lower-priority review. Not everything is a certain duplicate; some just deserve a second look.

FAQ

Common questions about duplicate detection

Duplicate detection is one of six. Explore the others.

See it work on your own expense data

A twenty-minute demo walks through how duplicate detection would work on your specific claim volume, vendor mix, and typical duplicate patterns.