To design a workflow process that is efficient, meaningful, and understandable it is important to properly consider how to compose the activities it will consist of. In this paper, guidelines are presented for this exact purpose. These focus on the elementary data-processing steps that are at the core of a workflow process. The guidelines help to determine the relative importance of these data-processing steps as well as their relatedness, such that activities can be composed in a fully automated manner. We implemented this approach in freely available software. A thorough evaluation that incorporates real-life workflow designs indicates that the use of these guidelines leads to activities that closely resemble those designed by experienced modelers. As such, the proposed guidelines provide a proper and automated alternative to what is otherwise a complex and time-consuming task.