This document summarizes a study on reducing failure analysis time for software testing in the automotive industry. The proposed approach clusters failing tests based on their coverage profiles and hypothesized faults, then selects representative tests from each cluster for analysts to examine. It was evaluated on 9 automotive components totaling 850 KLoC and over 7,000 tests, reducing failure analysis time by over 80% on average while maintaining high purity and accuracy in fault localization. The approach utilizes coverage profiles from software-in-the-loop testing to aid hardware-in-the-loop analysis and could help address challenges of failure analysis for automotive cyber-physical systems.