Representing concept lattices constructed from large contexts often results in heavy, complex diagrams that can be impractical to handle and, eventually, to make sense of. In this respect, many concepts could allegedly be dropped from the lattice without impairing its relevance towards a taxonomy description task at a certain level of detail. We propose a method where the notion of stability is introduced to select potentially more pertinent concepts. We present some theoretical properties of stability and discuss several use cases where taxonomy building is an issue.