In the social science literature, most data attribution is incomplete and does not include persistent identifiers (PIDs). Instead, authors mention data informally. Without explicit data citation, a publication cannot automatically or definitively link to a data source. The human effort required to find, interpret, and link opaque citations is costly and inefficient, so data use often goes untracked, and data creators go uncredited. In this paper, we outline the challenges specific to repositories, we categorize types of informal citation currently in use (based on our experience managing the ICPSR Bibliography of Data-related Literature, a continuously-updated database of more than 75,000 citations of works using data held in the ICPSR archive), and we offer suggestions for ameliorating the effort and expense of tracking data use, and for changing author behavior at the beginning of the publication process.