The Challenge

Our client faced challenges maintaining an accurate, comprehensive, and compliant database of all their customer addresses. The existing system suffered from inaccuracies and incompleteness, which incurred significant cost in lost and returned shipments. These inconsistent issues required a solution that could handle the large volume, velocity, and dynamic nature of address data.

Our Solution

Apption was chosen to design and implement a machine learning solution to deliver an accurate address database for shipment sequencing. This system would underpin also underpin new product opportunities for the company. Our solution handles the challenge adeptly, analyzing 40 million daily transactional shipment scans into a data aggregation framework. Apption built the solution using custom AI algorithms built on the Microsoft stack and the .NET framework, integrating seamlessly with the enterprise's existing systems.

The solution includes predictive learning models that classify raw data into a semantic network and compute a confidence score for each name and address using predictive clustering algorithms. The learning engine contains 3.5 billion connections, resulting in a customer database approximately 2TB in size.

The solution evolved over time, introducing features like business entity resolution, data enrichment, business and consumer mover predictions, zip/postal code targeting, and more.

Outcome

This project revolutionized the company’s shipment operations and Apption continues to support the system. It continues to validate crucial commercial data products for the digital services group, and is a critical part of the ongoing production systems.

Regarded as one of the largest innovations for predictive analytics in the corporation, this solution underscores Apption's capability as a key partner in developing data science and analytics solutions. It’s success is reflected in the improved accuracy, efficiency, and scalability of the shipment company’s operations.