Data on ethnic, racial and indigenous identity can help render certain minorities statistically visible, and expose potential discrimination and inequalities. This paper systematically reviews diversity data collection practices in OECD countries and selected key partners and identifies three common challenges: the legal treatment of 'sensitive' data and concerns around privacy; the use of different data sources for different policy purposes; and issues of comparability over time since identities are dynamic and multiple constructs. When relevant, recommendations and best practices to improve diversity data are put forward. These include: expanding the collection of data on ethnic and racial identities where legal frameworks permit; ensuring the representation of hard-to-reach populations such as indigenous communities; developing national diversity statistical standards to standardise information and allow linking data across sources; raising the timeliness and policy relevance of diversity data by including questions in both regular sample surveys and population censuses; and involving communities in the data collection process.