The goal of this paper is to reexamine assumptions about sketch-based interfaces for modeling in the context of designers' needs and practices. Research questions examine (a) the type of sketch support and (b) the timing of support. Both concepts try to determine when, what, why and how to augment design processes in a way that is useful to designers. Two experiments (one in architecture and one in product design) based on ergonomics theory are conducted and intend to question some of these assumptions. The Port Zeeland experiment examines how 20 novices perceive and copy a blurred architectural sketch, which provides clues for a sketch interpretation system. The “Tragere” experiment studies how 12 professional product designers, some of whom are “idea generators” and others “idea pursuers,” perceive, recognize, and handle a design sketch. The results take a designer's point of view in assessing the timing and value of sketch assistance in product design. The quantitative data analysis provides rich clues about when, why and how product sketches should be supported. The paper explores the strategies developed by designers to perceive and recognize graphical content and discusses the generation of three-dimensional volumes, the univocity state between sketches and three-dimensional models, and the treatment of features in freehand sketches. The paper concludes with observations on the timing and value of support, as first integrated in NEMo, a tool for early stage architectural design, and then in PEPS3, an early stage framework for product design.