We have been working with TEI for manuscript description for nearly ten years, creating large datasets and complex descriptions for catalogues and digital libraries. While we have benefitted greatly in terms of making individual collections or projects available online, we are unsure if we have fulfilled the potential which TEI offers for the creation of truly interoperable data, for shared development around standard datasets, and in the use of that data by researchers. In each of these areas, we feel that expectations are not necessarily matched by day-to-day experience. Efforts in community building around interfaces like the FIHRIST union catalogue have been largely successful. In the wider context, however, we have often encountered an initial resistance towards TEI. While it is recognised as a cutting-edge standard, academics, developers and curators often argue against the use of TEI owing to its complexity and the lack of standardised inputting tools. We organised a symposium to bring together librarians, developers and researchers to investigate these issues and discrepancies. What do these diverse groups really think of TEI as a standard for manuscript description? Why are some projects and institutions using it and not others? For those that are using it, what is their experience of creating interoperable data in TEI? And what is the experience of a researcher using this data to answer research questions? In this paper we will discuss the context of the Symposium and give a brief overview of the content of each session. But our main focus will be on the outcomes – what emerged which was useful or surprising, and how we might turn ideas and issues arising from the event into concrete actions.