Team coordination breakdowns (TCBs) generally reflect episodes of ineffective team functioning, resulting in suboptimal team performance. Computational identification of TCBs enables us to examine the underlying characteristics of suboptimal performance, and to potentially deliver real-time feedback to teams. Especially in time-critical crisis situations, such feedback can be invaluable. Previous studies found difficulties in distinguishing between coordination patterns that indicate TCBs, and patterns that might indicate other aspects of teamwork. Subsequently, we examined features capturing underlying characteristics of team coordination, based on multiple physiological signals and coordination measures, to identify TCBs. Our multi-methodological approach allowed us to identify features that are important for TCB identification. We also observed that distributions of feature data related and unrelated to TCBs were significantly different, indicating that the features captured underlying patterns in team coordination data. In addition, our results indicated that team performance, measured as goal achievement duration, is more severely compromised when TCBs last longer. By showing a relationship between TCBs and goal achievement duration, as well as understanding the key features of these TCBs, our study contributes to deepening our understanding of TCBs, and supporting effective team functioning and performance.