Email Risk Model Follow
What is Email Risk Model?
Email Risk Model is a global model that provides a comprehensive and holistic assessment of the riskiness of an email. This new model also introduces many new features that provide better risk predictability. Highlights of these new features include:
- Email tumbling detection
- Enhanced recognition and handling of disposable domains
- New Identity Graph and Network-based email features
What is the benefit of having Email Risk Model in Pro Insight?
For teams that utilize email as an identity verification attribute (for example, often looking to the Email First Seen signal in their manual reviews), the introduction of the Email Risk Model is an evolution that offers even better predictability of email risk.
How / where will Email Risk Model appear in Pro Insight?
Email Risk Model will return a risk-level flag (e.g., low, medium, high risk) as a signal underneath the email result column in Pro Insight.
There are only three possible results for Email Risk as a signal (low / medium / high):
How are the “thresholds” determined for an email to be considered low / medium / high risk?
Email Risk Model has gone through extensive testing as part of development, part of which includes evaluating multiple approaches to determine the best approach for maximizing performance and providing optimal thresholds for each country. As such, threshold determination has been optimized for use in Pro Insight, including support for a global customer base.
Will there be any specific signals returned that identify if an email is tumbled, autogenerated, or has a disposable domain?
There are no specific signals that identify these types of fraud. Detection of these fraud types is built into the fraud model. For example, if the model identifies an email address has a disposable email domain, it will return a higher score accordingly. Email Risk Model in Pro Insight will show up as a single signal (Email Risk) within the Email result column categorizing risk level as a flag (low / medium / high).
In other words, Email Risk Model will appear as a signal indicating email risk is either low, medium, or high:
Is email address the only required input?
Yes. However, we strongly recommend providing the country of transaction or identity where possible.
This is because the Email Risk Score can be different depending on the country of interaction. The model will perform best when this information is included in the query.
Email Risk Model will be able to identify country through any of the following provided inputs:
- Physical address (primary or secondary)
- Phone number
- IP Address
Will Email Risk Model replace the Email First Seen signal?
The Email First Seen (EFS) signal is also a feature in Email Risk Model, which means it contributes to the model’s assessment of remail risk. As a result, we would expect (and have so far seen in testing) that Email Risk Model provides better risk predictions.
However, there are currently no plans to retire the Email First Seen signal from view in Pro Insight, and manual review agents can continue to reference EFS in addition to the Email Risk signal.
Is the Email First Seen Signal in Email Risk Model?
Yes. Email First Seen (EFS) is one of many features in Email Risk Model.
How can I share feedback on this model in Pro Insight?
Feedback is always appreciated! Additionally, if the new Email Risk signal has helped your team act more efficiently and effectively, we would love to know more. Please share feedback with your client success manager or reach out to us at firstname.lastname@example.org.