This project came about as a new feature to use machine learning to assist with electronic correlations (combining electronic payments with emailed remittances). The process was very manual originally taking a minimum of 5 clicks to submit each correlation and usually more to find the right payments and remittances to match. We reduced it to 3 total clicks for automatically matching everything the algorithm could identify as potential matches, saving hours of work.

The solution was composed of the following:

• A new screen for reviewing and approving the automatic matches with a table format
• A new screen for reviewing and approving the automatic matches with a grid format
• Updates to the Correlation History screen to allow users to remove automated mappings manually.
• Updates to the Correlation Review screen (the original manual screen) to allow users to identify which payments and remittances were auto-matched by the algorithm.

This project was user tested and went through 4 rounds of design iterations (listen, user test, design) before the final version was handed over to development to build and release.

© 2016 Billtrust. All Rights Reserved.

• Client Billtrust
• Created by Nate Girard
• Completed in 2015
• Skills Prototyping / User Testing