Techniques described herein include determining, maintaining, and applying compound service performance metrics, based on data metrics from a plurality of different services. Service-specific data metrics may be received from a plurality of different communication services offered by a service provider, for example, Internet service, voice service, video service, SMS service, etc. Different combinations, relationships, and weighting factors for the data metrics may be defined and stored for each compound performance metric. Compound performance metrics may be defined, including for example, compound customer sentiment metrics, compound customer value metrics, and/or compound customer resource usage metrics. In some cases, machine-learning and/or analytics may be performed using service-specific data metrics and corresponding customer actions, in order to determine correlations between particular combinations of data metrics and customer actions.