Conceptual modeling practice in agile development methodologies an exploratory study into the underlying assumptions of conceptual modeling practice

Conceptual modeling is the practice of formally describing a real-world domain to enable better understanding and communication among stakeholders. It provides a basis for design during the development of information systems. In a sequential approach to system development projects (e.g., via traditional waterfall methodologies), a priori specification of system requirements via complete, clear conceptual modeling scripts is deemed essential to the success of a project. In the current practice of system development, however, non-sequential methodologies such as agile methodologies are gaining increasing prominence. Conceptual modeling supposedly does not have a primary role to play in these methodologies. Nonetheless, use of these methodologies seems not to have increased failure rates in system development projects, even though they are deemed to downplay the role of conceptual modeling. My research uses an ontological perspective on conceptual modeling to explore the practice of system development when agile methodologies are used. I conducted an interpretive field study involving semi-structured interviews with eight highly experienced practitioners to explore the context and methods of conceptual modeling in agile settings. I provide an explanation of the anomalies that exist in the perceived role of conceptual modeling when using agile methodologies, compared to waterfall methodologies. Based on the findings of my study, I have concluded that contrary to much current rhetoric, the practice of conceptual modeling is not becoming obsolete in agile methodologies. Rather, its importance is growing. This outcome is occurring with agile methodologies for two reasons. First, information is increasingly recognised as a concrete asset that brings value to organisations. Therefore, procedures such as conceptual modeling that enable information about domain semantics to be extracted more easily are gaining significance. Second, advances in technological infrastructures have enabled practitioners to focus more on a domain’s semantics as many system implementation details are becoming standardised. In spite of growing significance of conceptual modeling, the findings of my study indicate that fundamental differences exist between conceptual modeling practice in agile methodologies and traditional sequential methodologies. The differences are twofold. First, agile methodologies differ from sequential methodologies in terms of the level of granularity of the conceptual models used to represent domain semantics. In agile methodologies, detailed, a priori specifications of domain semantics are not needed. Instead, domains are often represented only in terms of their main subject matter. In this regard, conceptual models in agile methodologies are deemed to be coarse-grained representations of domains (compared to their sequential counterparts that attempt to provide complete, finely grained representations of domains). Second, the results of my study show that the practice of conceptual modeling is influenced by the context and overall objective of the information system for which the conceptual models are developed. While formalisation of conceptual modeling in traditional methodologies seems to be unaffected by different types of information systems that exist, a theme of System Taxonomy emerged through the analysis of data in my study. This theme indicates that domain uncertainty and volatility, as well as the overall objectives of different information systems, influence the practice of conceptual modeling in agile system development. My study makes contributions to the body of knowledge through development of new concepts and provision of rich insights about real-life practices of conceptual modeling. It also expands the boundaries of current theories about conceptual modeling (initially developed in sequential settings) by showing their relevance, at least in part, in non-sequential settings.