We look at how to design a comparative interpretive project and tackle the perennial case selection question. The problem here is one of justifying unorthodox comparison: a lot has been written on comparative case selection from a naturalist perspective, but this language is often an uncomfortable fit for interpretive projects. We argue that case selection is not something that is designed into a project from inception. For interpretive research, it changes as we go. We therefore suggest different strategies of case selection for different phases of a comparative interpretive project. We identify rules of thumb to guide design choices as the project or programme evolves.