The CARARE metadata schema was designed to support harvesting archaeology and architecture. Monuments and buildings are complex, multi-layered objects with dynamic lives. A wealth of digital information objects relating to different aspects of the tangible and intangible heritage exist - from texts, plans, drawings and images, to folk memories, diaries, stories, news papers and traditional skills. The digital heritage landscape is now going beyond metadata for individual objects instead seeking to support associations between objects, places, people and events, integrating information sources and providing for narratives. In developing version 3 of the CARARE metadata schema the authors seek to increase the support for linked data and contextualisation, and to support new uses of the schema particularly as a data capture format. The paper describes the development methodology, use cases, design challenges and considerations made in defining the new version, and prototyping of the schema. Semantic Omeka is being used to test the data capture use case.