Case Study: a major Canadian bank improves fraud detection speed and accuracy with CGI machine learning

Major canadian bank cuts false positives by 20% with CGI

CGI assisted a major Canadian bank with the challenge of improving its fraud detection capabilities. The bank's existing rule-based system was struggling with false positives and false negatives, and its batch processing approach delayed fraud identification. These limitations forced a trade-off between customer experience and financial protection.

The solution involved CGI implementing advanced data engineering to enable near real-time processing and then introducing supervised, semi-supervised, and unsupervised machine learning models to augment the rule-based engine. This allowed for a more agile and accurate detection system. As a result, CGI helped the bank reduce false positives for customers by 20%, detect fraud faster, and improve overall system efficiency and security.