A sensor-based project management process, which uses continuous sensing data of face-to-face communication, was developed for integration into current project management processes. To establish a practical process, a sensing system was applied in two software-development projects involving 123 and 65 employees, respectively, to analyze the relation between work performance and behavioral patterns and investigate the use of sensor data. It was found that a factor defined as "communication richness," which refers to the amount of communication, correlates with employee performance (job evaluation) and was common in both projects, while other factors, such as "workload," were found in just one of the projects. Developers' quality of development (low bug occurrence) was also investigated in one of the projects and "communication richness" was found as a factor of high development quality. As a result of this analysis, we propose a four-step sensor-based project management process, which consists of analysis, monitoring, inspection, and action, and evaluated its effectiveness. Through monitoring, it was estimated that some "unplanned" events, such as changing specifications and problem solving during a project, could be systematically identified. Cohesion of a network was systematically increased using a recommendation of communication, called WorkX, which involves micro rotating of discussion members based on network topology.