I argue that conceptual engineering, on the understanding I favour, is the identification and naming of the properties we need for this or that worthwhile purpose, independently of whether they are properties named in extant theories. So understood, it is obvious that conceptual engineering should be supported, and I give some non-controversial examples. I discuss the connection between our description of conceptual engineering in terms of properties and words, and the more usual one in terms of concepts. The implications of conceptual engineering for a number of live issues in analytical philosophy—knowledge, Swampman, narrow content, proper names—are discussed. I conclude by noting that our defence of conceptual engineering does not imply that traditional conceptual analysis has no worthwhile role to play but does imply that its role needs a certain amount of rethinking. I illustrate this with the example of personal identity.