TrustGraph detection categories

The kinds of fraud that TrustGraph is designed to flag.

Each category describes a different way fraud shows up across a network of institutions.

TrustGraph organizes its fraud signals into five detection categories: Suspected Repeat Fraudster, First-Party Fraud, Unusual Activity, Velocity and Excessiveness, and Fraud Rings. A detection category is a named type of risk that a signal can point to.

The five categories are:

• Suspected Repeat Fraudster, which points to an identity that has appeared as a problem elsewhere.
• First-Party Fraud, which points to users who act anomalously, as reported by an institution, such as rewards abuse, account selling, or mule activity.
• Unusual Activity, which points to behavior that departs from what is expected.
• Velocity and Excessiveness, which points to too many attempts or too much activity in a short time.
• Fraud Rings, which points to coordinated groups of linked identities.

A category works within a general flow:

1. An institution runs a check as part of its identity flow.
2. TrustGraph evaluates the available signals against the categories.
3. It returns a result for each category, and the institution combines it with its own rules.
4. The institution decides whether to approve, decline, or send the case for review.

Which categories a customer sees can depend on how TrustGraph is deployed and released, so the list is best read as the range of risks the product addresses.

The categories sit next to terms that are easy to mix up:

• First-party fraud is a category and also a glossary term in its own right.
• Fraud ring is a coordinated group, and also a glossary term.
• Velocity checks are rules that count how fast something happens, which relates to the velocity category.