Designing and building a knowledge management system involves ensuring that the right facts can be called upon to answer the question at hand, and coordinating a number of disparate resources. Modelling such resources can be problematical as there is currently no formalism that can represent the nature of the data- seeking process at a conceptual level. We introduce the functional entity (FE), an encapsulated data resource that acts as a question-answering system, and identify nine different functional entities based on three main types of question-answer entailment: instance-dominant, value-dominant, and connection-dominant. We use functional entities to develop a generalisation of the Entity-Relationship Diagram (ERD), the Functional-Entity Relationship Diagram (FERD), which can be used for high level conceptual modelling of heterogeneous KM systems, and illustrate its use with a case study. We also show how the establishment of standard types of functional entity motivates practical guidelines for the conversion of the design to the implementation level.