The surge in usage of mobile applications generates a massive volume of traffic data exhibiting unique dynamics that are hard to unravel. In this work, we leverage factor analysis to pin down recurrent patterns of mobile traffic over the three dimensions of space, time and services in multi-city measurements of unprecedented resolution. We link the revealed structures of real-world mobile demands to urban fabrics, i.e., the combination of infrastructures and social characteristics that determine the functionality of urban territory, hence establishing connections between specific city landscapes and the mobile application consumption they create. Our study provides a new understanding about the diversity of mobile service dynamics in metropolitan areas, including insights on how economic status drives the adoption of specific applications, how residential versus commercial areas create a dichotomy in applications usage, how private and public transport drive surges in the prevalence of different sets of applications or how nightlife or university studies stimulate the utilization of specific classes of services.