Strong claims are made by both higher education policymakers and institutional leaders about the advantages that can be brought about through 'datafication' (i.e. the quantification of human activities through digital means). Nevertheless, we know relatively little about how university cultures and processes may impinge on the rolling out of data practices, particularly in relation to administrative data (rather than that collected through learning analytics and other forms of user-generated data). This article draws on a particular case study--of attempts to widen participation to 'sandwich courses' in English higher education institutions--to argue that, in some areas of activity at least, institutional constraints serve to limit datafication substantially, including insufficient time and infrastructure, the prioritisation of data required for other purposes (typically national assessment exercises) and the role of institutional cultures. Cognisant of various problems associated with quantification, the article nevertheless contends that the barriers to the effective use of metrics identified in this research may have a direct impact on institutions' capacity to recognise and address social inequalities.