Object identification is a fundamental aspect for efficient communication between two agents (humans or software) operating in the same context; requiring agents to have a shared understanding of the objects' characteristics and of their current context. In case these agents do not have the same observing capabilities, establishing a shared understanding can become significantly challenging, and could lead to communication failure. In this article, we explore the design space of automatic identification systems, highlighting the absence of a formal model of (automatic) object identification that help agents choose optimal discriminative characteristics of an object to differentiate it from other objects in a given context. Our objective is to deepen the understanding of automatic identification approaches and systems. Hence, we propose a model that supports designers of (automatic) identification systems in pinpointing discriminative characteristics of an object that are pertinent to consider in a given use case. This model enables structured analysis of the compatibility of diverse (multimodal) identification systems, promoting effective communication among agents, and opens a path for systems to automatically select discriminative characteristics at run time. Our proposed model considers endogenous and exogenous object characteristics and their variations over time. It also includes the perceptual subjectivity of these characteristics in identification systems. To validate our model, we implement scene understanding systems that showcase, in various scenarios, the discriminatory power of object characteristics, and demonstrate the role of our proposed model in enabling effective communication between two systems (agents) with differing capabilities.