A rapid changeover capability is widely acknowledged as an essential prerequisite to flexible, responsive small batch manufacturing. Its importance in mass customization is recognized, where minimal losses need to be incurred as manufacture switches between differing products. Retrospective improvement of existing changeover practice is often undertaken, arising from pressure to respond better to customer demands, wherein improvement personnel frequently engage Shigeo Shingo's Single Minute Exchange of Die (SMED) methodology to achieve better performance. Irrespective of the improvement methodology that is employed this paper assesses two fundamental mechanisms by which better changeovers might be achieved. First, improvement can occur by altering when tasks are conducted. Better allocation of tasks to the resources necessary to conduct them is sought, where the tasks themselves remain essentially unchanged. The second mechanism is to seek structural change to existing tasks, thereby intrinsically enabling them to be completed more quickly. These two mechanisms are described in relation to use of the SMED methodology, where it is argued that, by reinterpreting Shingo's work, greater clarity of potential improvement options can be gained.