This methodological contribution points out some of the problems associated with so-called parametric approaches to the study of morphosyntactic variation in the Bantu languages, and proposes an alternative, scenario-driven method for discovering patterns of variation and for organising linguistic data in a manageable and meaningful way.