An enhanced version of the public REF 2021 submissions dataset, produced by Jisc, which contains metadata for all outputs submitted to the exercise. Metadata has been cleaned and new fields have been added to increase the potential for analytical purposes, e.g. by identifying where the publisher exists as an imprint of a larger parent company. For its own analytical purposes, Jisc focused on long-form output types (books and parts of books), but cleaning measures were performed on the entire dataset, uploaded here.