Mobile devices have conquered our daily life. They are carried and used at almost all times and in any situation sometimes even against legal restriction - e.g. in the car. Services running on these devices (e.g. email, text messages, etc.) include graphical (GUI) and voice interface (VUI) causing visual distraction for the driver although they could solely be operated by speech in- and output. As a result services should adapt their interfaces due to the specific usage scenario. The aim of this work is therefore to develop design recommendations based on a cognitive model for the voice user interface taking the particularities of the specific scenario (e.g. reduction of off-road eye gazes) into account. We assume that distraction is mainly due to an increased effort and a decreased expectancy of the VUI compared to the GUI. Design recommendations derived from these considerations will be described by means of a concrete example.