Scoring and Decisioning

The Scoring Engine is SEON's main tool for preventing fraud by examining transaction data. It works by applying rules and using machine learning models to measure risk and assign a fraud score to each transaction. Based on the score, a transaction is either approved, flagged for review, or declined. This system helps businesses make better decisions with minimal manual effort.

How decisioning works

SEON evaluates transaction data by adding extra information from the API request and analyzing it for risk signals. A fraud score is then calculated based on this analysis. If the score is low, the transaction is approved. If moderate, it requires manual review. If high, it is automatically declined. Each rule applied contributes to the total fraud score, and all rule evaluations are included in the API response and event details for full transparency.

Rules

SEON uses rules to decide whether a transaction is risky. Default rules are built-in and target common fraud patterns but cannot be changed. Custom rules can be created by businesses to adjust scores or directly approve, review, or decline transactions based on specific conditions. These rules can raise or lower a fraud score depending on risk factors observed in the transaction data.

Machine learning

SEON includes two machine learning models to improve fraud detection. Whitebox machine learning creates clear, explainable rules using historical data and provides accuracy metrics. Blackbox machine learning generates a fraud probability score from 0 to 100 by identifying complex patterns in transaction data that standard rules might miss. Both models help detect fraud more effectively over time as they continuously learn from labeled transaction data.

List management

Lists play a vital role in controlling how transactions are handled. The blacklist blocks transactions with flagged data points by automatically declining them. The whitelist allows transactions from trusted sources by approving them automatically. Custom lists help businesses monitor certain data points for ongoing analysis without immediately affecting the fraud score. These lists give businesses more control over their fraud prevention strategies.

This structured decisioning process ensures effective fraud prevention while giving businesses the flexibility to customize their risk management strategies.