Under the hood of Pro Insight is the Identity Engine, a key competitive differentiator across all Ekata products. But what exactly is it? A combination of sophisticated data science and machine learning, the Identity Engine combines two powerful and proprietary datasets. Our Identity Graph and Identity Network provide unique and valuable insights that allow businesses to accurately make risk decisions about their customers and improve the manual review process.
Machine learning models can be powerful tools that support manual review agent teams in confidently assessing fraud risk at scale. However, they can, at times, also seem like a bit of a black box. How did a model arrive at this risk score? What reason codes had the greatest impact? And, more practically still, what particular elements from the Identity Graph and Network can a manual review agent leverage as justification for approving a low-risk customer or rejecting a high-risk customer?
That’s where reason codes come into play. They help manual review agents peek behind the curtain of Ekata’s complex machine-learning models.
Using Reason Codes to Understand the Identity Risk Score
Reason codes, the positive and negative signals located on each side of the Identity Risk Score, provide insights into the Identity Graph elements driving a score low or high. Without strong signals either increasing or decreasing the estimated risk, a transaction is more likely to appear uncertain. Reason codes help manual review agents understand the context and logic behind the score. As dynamic signals, reason codes draw from all transaction inputs (the five core digital and traditional data attributes of name, email, phone, address, and IP) to display the most impactful signals contributing to the Identity Risk Score.
Reason codes are weighted by strength (represented by a strength indicator with a scale of 1-5) and listed in order of importance, with the strongest signals listed first. This matters because understanding which identity elements positively or negatively contribute to a low/high-risk score enables a manual review agent to focus on the identity elements that seem most suspicious. The importance of a reason code will depend on the context it is being analyzed in. For example, in cases of first party fraud versus account takeover fraud.
In the case of the former, positive signals that indicate an exact or close match to the phone subscriber name, as well as a valid primary address, help confirm it is an authentic transaction. This allows companies to fight against chargebacks. With the latter, negative signals that indicate IP address is risky, lack of a registered name for the email and absence of a phone subscriber name helps categorize it as a potential account takeover attempt and therefore should be rejected.
Save Time on Manual Review – and More!
While it is important to consider a variety of signals behind an Identity Risk Score, it is also important for manual review agents to understand and weigh the relative importance of each signal to assess the level of risk appropriately and accurately. An important part of getting the most out of Pro Insight is deciding the rules and risk thresholds appropriate for your manual review team. This is necessary so organizations can optimize resource spend and efforts on mitigating the most significant risks; balancing risk and reward for maximum revenue. Reason codes provide some visibility into the models powering our Identity Engine and Pro Insight and help manual review agents justifiably make decisions faster and more effectively. The end result? More good customers and increased revenue!
Interested in learning more? Empower your manual review teams to help your business grow more efficiently while balancing bad and good transactions. Contact an Ekata fraud expert today to schedule a Pro Insight demo and/or request a free Pro Insight trial.