In team sports, the group performance is driven largely by the network of interactions among individual team members having non-exchangeable roles, and how these interactions change under different conditions and in the presence of an adversary. Classical approaches used in psychometrics and social network analysis tend to rely on unrealistic assumptions such as independence or exchangeability and to only consider a single network instead of multiple networks cross-classified by categorical variables. This paper explores the relationship between teams’ passing tactics and a number of situational factors in soccer games. We propose a multiresolution data representation framework for spatial passing networks, and an exploratory analysis tool based on tensor factorization, which is able to produce coarse-to-fine network motifs and associate them to the situations teams encounter. We evaluate our approach using detailed on-ball events data collected from the FIFA 2018 World Cup by StatsBomb.