Dynamic, evolving systems pose new challenges from the point of view of Quality of Service (QoS) analysis, calling for techniques able to combine traditional offline methods with new ones applied at run-time. Tracking the evolution and updating the assessment consistently with such system evolution require not only advanced analysis methods, but also appropriate metrics well representative of QoS properties in the addressed context. The ongoing European project Connect addresses systems evolution, and aims at bridging technological gaps arising from heterogeneity of networked systems, by synthesising on-the-fly interoperability connectors. Moving from such ambitious goal, in this paper we present a metrics framework, whereby classical dependability/ QoS metrics can be refined and combined to characterise Connect applications and to support their monitoring and analysis.