This article reports on action research that was undertaken while teaching a mathematics education subject to a series of cohorts of final-year primary pre-service teachers at a university in Australia. My aim was to increase the understanding of an assessment task, as well as its intrinsic value. Each action research cycle involved the introduction of a variation in the assessment task's instructions, a process informed by variation theory. The variations were introduced through cycles of planning, action, data collection, and reflective analysis related to clarification of the assessment task. The two theories (variation theory and action research) proved to be mutually supportive. Cycles of action research served well to identify the critical features that needed to be the focus of small variations in my presentation of the assessment task. Variation theory provided guidance as to a most effective way of highlighting these features, especially with regard to foregrounding varied requirements of the task.