The ethics of big data and AI have become the object of much public debate. Technology firms around the world have set up ethics committees and review processes, which differ widely in their organisation and practice. In this paper we interrogate these processes and the rhetoric of firm-level data ethics. Using interviews with industry, activists and scholars and observation of public discussions, we ask how firms conceptualise the purposes and functions of data ethics, and how this relates to core business priorities. We find considerable variation between firms in the way they use ethics. We compare strategies and rhetoric to understand how commercial data ethics is constructed, its political and strategic dimensions, and its relationship to data ethics more broadly.