What is a fraud score?
A fraud score is the computed result of Ravelin’s machine learning analysis that that takes into account over 600 different fraud signals.
The fraud score is a probability of how likely a customer is to be a fraudster. For example, a person with a fraud score of 30 is 30% likely to be a fraudster.
Where can I view a customer’s fraud score in Ravelin?
You can view a customer’s score on the customer lists page or a customer’s profile. This is a feature that is only available to our Enterprise clients.
There are 3 places where you can see the fraud score
- On the customers list page
- The customer list on the client profile page
- In the Overview, on the client profile page
How do I use a fraud score?
With Fraud scores you can prioritise what actions you should take on a customer. You can determine your business’ score thresholds based on your tolerance for risk and bandwidth for manual review.
- Genuine scores: These users are likely good and don’t require review. Ravelin will tell your application to ALLOW the customer to place an order.
- Review scores: These customers need review. They may be good or bad and need you to dive into their details to make an educated decision. Ravelin will tell your application to REVIEW the customer’s order.
- Fraudster scores: These customers are likely bad and should be automatically blocked. Ravelin will tell your application to PREVENT the customer from placing an order.
Note: If you would like to update your scores thresholds, please contact firstname.lastname@example.org.
What are fraud score contributions?
We have grouped contributions so it's easy to understand why a customer has a given score.
- Identity - Looks at the customer information, email, phone number...
- Orders - Looks at customer's orders, type of items, order and transaction velocity.
- Payment Methods - Looks at the different payment methods the customer has used,
- Locations - Looks at the different locations from billing address, delivery address to the customers location.
- Networks - Looks at the customer network, number of linked account, devices.
For each group you can see the most important contributions that were used to determine the fraud score.
For innocent users with low scores there could be cases with no contributions because they don't exhibit fraudulent behaviour.