Glossary term
Fraud Detection
Subject: Financial Crime, Fraud
Fraud detection is the process of identifying suspected fraudulent activity after or while it happens, using rules, statistical models, data analysis, alerts and human reports, so that cases can be investigated.
Explanation
Prevention tries to stop fraud occurring. Detection accepts that some will get through and aims to spot it quickly. Banks run rules on transactions, such as a card used in two countries within an hour, and machine-learning models that score each payment for risk. Retailers and insurers compare claims or orders against known patterns and linked identities. Internal teams use data analytics to find duplicate invoices, ghost employees or round-sum payments. Human sources matter just as much: customer complaints, staff tips and audit findings. A good system balances catching real fraud against false alerts that annoy genuine customers and swamp analysts. Alerts feed investigators, who decide whether to block, report or recover funds. Each hour saved between the first suspicious transaction and the first alert is money that can still be stopped, and evidence that is still fresh.
Example
A bank's monitoring system flags a customer account that receives a payment, then sends 95% of it to a new payee within minutes. The score is high because the device is also new. An analyst calls the customer, who confirms she was pressured by a caller to move the funds. The bank stops the transfer and records the details to improve its rules.
Not to be confused with
- Fraud prevention
- Prevention puts barriers in the way before an attempt succeeds. Detection finds attempts and losses that get past those barriers.