One of the key activities that are needed to construct a quality service-oriented solution is the identification of its architectural elements with the right granularity. The selection of an appropriate method for identification of services from business models of an enterprise is thus quite crucial to the success of any service-oriented solution for that enterprise. Existing methods for service identification ignore required performance metrics and semantics integrity of business elements; more importantly, they focus on entity-based services while ignoring process based ones. This paper proposes a new process for identification and specification of enterprise software services and their architectural elements. A novel clustering technique, named elementary business process and business entity affinity analysis technique (EEAT), is introduced for identifying candidate architectural elements. This technique identifies each candidate service with the right granularity, while satisfying low coupling, high cohesion, and low reuse cost principles for reusable software services.