This paper discusses the common challenges and considerations associated with the development of custom taxonomies for describing people along the lines of race, gender, sexual orientation, nationality and other identity facets. The paper recognises that there are inherent and divisive problems in reducing complex human identities to discrete categories, and that terminology is always changing. Nevertheless, in the absence of a public vocabulary that addresses all facets of identity in a respectful and accurate way, custom taxonomies remain necessary. This paper explores the strengths and weaknesses of existing vocabularies, supporting this analysis with real-world examples, including a discussion of the LGBTQ+ vocabulary, Homosaurus and alternative terminology sources. The paper then proposes a flexible process for researching and developing custom taxonomies that draws on the strengths of existing vocabularies, knowledge of one’s content and users, professional best practice literature, journalistic style guides, and first-person consultations. The paper argues that custom identity taxonomies are a way of showing respect for the people represented in one’s digital asset management system and the people who use it. Thus, bringing the values of diversity, equity and inclusion into one’s metadata are a small way of expressing and fostering a community of care.