This talk is about the methodology of “social metaphysics” in the sense of socially significant metaphysics. This is not quite the same as the metaphysics of social entities or kinds. For example, whether robots can be conscious is a question of “social metaphysics” in the relevant sense, although a robot is presumably no more a “social” entity than a human individual. The question is: should inquiry into social(ly significant) metaphysics be untainted by considering the potential upshots for social practices? I call the approach that answers this question affirmatively the “metaphysics-first” approach. According to the contrasting, “ethics-first” approach, we should consider the practical consequences of the metaphysical verdict (for example, of recognizing robot consciousness or group agency or of accepting a certain account of gender) in the course of reaching the metaphysical verdict. I also distinguish between second-order metaphysics-first and ethics-first approaches to defending either of the first-order approaches. On the second-order metaphysics-first approach, the relevant grounds concern metaphysics: we should take either the ethics-first or the metaphysics-first approach because this is required for doing metaphysics well, for getting the metaphysics right. On the contrasting, second-order ethics-first approach, either of the first-order approaches is motivated by ethical considerations. The aim of the talk is to further flesh out these distinctions and to defend the ethics-first approach to social metaphysics, by also taking a second-order ethics-first approach to the defence.