Fraud detection

Fraud detection is the identification of fraudulent activity, either while it is under way or after it has happened, by finding signals in transaction, account and device data that separate it from legitimate behaviour. It judges events; it does not by itself determine what happens to them.

In more detail

Detection and prevention are often used interchangeably, but they answer different questions. Detection asks whether this activity looks fraudulent. Prevention asks what the organisation does about it. A detection method can be excellent and the surrounding process still let the loss through, if nobody acts on the alert in time.

Detection is also not synonymous with machine learning. Hand-written rules, statistical baselines and learned models are all detection methods, and mature programmes usually combine them because each misses cases the others catch.

In a working system, each event is evaluated against rules, a score is calculated, and anything above the review band is routed to an analyst queue.

In a working system, one engine evaluates each event against rules, calculates a score, and routes anything above the review band to an analyst queue.

Example

A card is used for a small online purchase in one country and, forty minutes later, for a larger purchase in another. Neither purchase is unusual alone. The pair is flagged because the card cannot plausibly be in both places.

Related capability

Fraud prevention and detection on the platform

Related terms

Discuss your transaction monitoring requirements

Tell us which transaction flows you need to monitor, which systems Akoben would integrate with, and how your fraud team works today.

Request a demo