Abstract. This paper presents a methodology for conceptual modeling which is based on a new modeling primitive, the niche, and associated constructs granularity and reconciliation. A niche is an environment where entities interact for a specific purpose, playing specific roles, and according to the norms and constraints of that environment. Granularity refers to the relative level of power or influence of an entity within a niche. Reconciliation is a relationship from N entities onto one reconciled entity, and represents explicitly a situation where two or more different perspectives of the same entity have been reconciled, by negotiation, into a single consensus view. The methodology we propose provides a systematic method of designing conceptual models along with a process for normalising inappropriate relationships. Normalising is a prescriptive process for identifying and remedying inconsistencies within a model based on granularities. Drawing on a number of case studies, we show how niches and granularity make complexity easier to manage, highlight inaccuracies in a model, identify opportunities for achieving project goals, and reduce semantic heterogeneity.