Reason Codes

Scores and Reason Codes

Fraud and identity risk systems typically return two primary types of outputs:

• Scores
• Reason Codes

These outputs are designed to help organizations make risk-based decisions while maintaining appropriate model transparency.

Example Response

Below is a simplified example of a successful response from a risk evaluation service:

{
  "referenceId": "SAMPLE_ID",
  "risk": {
    "reasonCodes": ["R101"],
    "scores": [
      {
        "name": "risk_model",
        "version": "1.0",
        "score": 0.29
      }
    ]
  }
}

Scores

The scores object contains:

• A numeric risk score
• Model metadata such as name and version

The score represents the model's overall assessment of risk for the transaction.

How the score is interpreted (for example, what qualifies as low or high risk) depends on your organization's configured thresholds and risk policies.

Reason Codes

The reasonCodes array provides high-level indicators of factors that influenced the score.

Reason codes:

• Offer limited transparency into contributing signals
• Represent a subset of evaluated model inputs
• Do not expose detailed feature logic or model design

They are intended to support operational review processes, particularly for transactions that fall near decision thresholds.

Important Considerations

• Risk models evaluate many signals to generate a score.
• Not all contributing factors are exposed.
• Reason codes are informational only and have already been factored into the final score.
• They should not be interpreted as standalone risk rules.

For a complete and current list of supported reason codes, refer to the Reason Code reference documentation in your API materials.