Table of contents
What do I do if a customer contacts me regarding a declined order?
There are multiple types of declined orders. You will want to communicate with customers differently based on this:
- "Fraudulent" Declines are our highest level of fraud decline. We strongly advise against engaging with the buyer, as there is a high likelihood that they are a fraudster.
- "High-Risk / Considerable Risk" Declines are more borderline. Therefore, you may decide to further discuss the order with the customer (i.e., verify the order details). If you are confident the order is safe, you can decide to fulfill it without Riskified’s guarantee.
- "Business" Declines are triggered by rules that are based on the business policies of your company. For example, some merchants do not want any orders to be sent to reshippers. Therefore, Riskified will set up a business rule that will automatically "business" declined any reshipper orders. If you need assistance with your business rules, please reach out to your Account Manager or firstname.lastname@example.org
Where can I learn more about why an order was declined?
Riskified developed "Decision Insights" to give merchants more information into declined orders. Decision Insights are available for every declined order inside the Control Center and are provided to give merchants a deeper understanding of why that order was declined.
Declined orders can be classified as "Fraudulent," "High Risk" or "Considerable Risk." These classifications indicate how conclusive Riskified’s evaluation was for that order.
Another useful feature within the Control Center is the radar chart display. This chart shows the effect that each of five major risk categories had on Riskified’s decision. The larger the category appears on the chart, the more significant it was in the decision process.
How does Riskified decide if an order is fraudulent?
Riskified reviews every submitted order using machine learning models and proprietary analytical tools. Our algorithms consider data including customer details, proxy detection, IP and geolocation, device and browser fingerprinting, and behavioral data. This information is checked against millions of other orders before a risk profile is determined.