The Transaction scoring API documentation describes the endpoints and response objects that are provided by the scoring service. This article also discusses the underlying architecture and how the scoring engine works to identify suspicious patterns of behavior in your app’s users.

This API returns a fraud score that indicates the likelihood of fraudulent activity for a given transaction. This score is calculated based on an evaluation of the payment details and user data by several fraud detection services. It uses a weighted average, where services and risk signals with greater impact have a larger effect on the overall score. The values of the rules triggered by a transaction will also contribute to the overall fraud score. The Applied Rules widget provides a breakdown of these rules, including the decision type assigned (APPROVE, REVIEW, DECLINE).

Transaction Scoring API Documentation: How to Detect Fraud

The response object contains several important fields. The risk decision and a risk elements object are key fields that provide the fraud score and other information for each detected pattern.

A transaction must pass a certain minimum number of parameters for this service to evaluate it and return a result. Some of these variables are required while others can be optional. An example of optional transaction data is the billing or shipping address. While passing these variables can help the service detect abuse, only a full lookup with our dedicated Email Verification API can provide more comprehensive fraud prevention.

Some of these variables are checked using standard operators like is equal to and is not equal to. For instance, the check for a device with 0% battery will evaluate a value based on whether or not it matches an existing one in your system.