What should conceptual engineering be all about?

Conceptual engineering is commonly characterized as the method for assessing and improving our representational devices. Little has been said, however, on how best to construe these representational devices—in other words, on what conceptual engineering should be all about. This paper tackles this problem with a basic strategy: First, by presenting a taxonomy of the different possible subject matters for conceptual engineering; then, by comparatively assessing them and selecting the most conducive one with a view to making conceptual engineering an actionable method, that is, a method that can be applied effectively and consistently to specific case studies. The outcome is that conceptual engineering should be all about concepts on pain of pragmatic inconsistencies otherwise.