While historical data isn’t required, it’s recommended. Building an effective model requires a complete, robust, and accurate database with past transactions and their outcomes (if available).
It also requires any further information that could be used in the authentication process. Our intelligence team provides support for this extraction process and may ask for a pilot sample before the final extraction.
We automatically provide new versions of the authentication score free of charge. The historical database must be sufficiently large to enable the extraction of a statistical mass that lets us develop effective models. For the best results, we recommend our clients send the last six months of transactions, including:
- Order number
- Order date
- Customer name
- Receiver name
- Customer email address
- User registration date
- Email address that referred the user, if available
- Order IP address
- Full billing address
- Full shipping address
- Customer phone number 1
- Customer phone number 2, if available
- Order amount
- Items being acquired and individual value of each item
- Order channel
- First six digits of credit card
- Last four digits of credit card
- Credit card expiration date
- Cardholder name
- Status indicating whether the order was auto-approved, approved, auto-declined, declined for fraud-related reasons or declined for other reasons
- Flag identifying whether an order originated a fraudulent chargeback
- Chargeback notification date, if available
Businesses can transmit this data to ClearSale in several ways, including:
- API integration
- CSV files, to be transmitted through an SFTP connection
- Database dump, to be transmitted through an SFTP connection
Related Topics
Ecommerce Platforms and Integrations
Related Resources
ClearSale Ecommerce Fraud Protection Solutions for Small-Medium Businesses