In recent years, there has been a growing interest in the modelling of crime commission processes, in particular crime scripting, in physical and cyber spaces. This article aims to demonstrate the limits of unstructured scripting approaches, and advocates the development of more systematic techniques. For this, we examined the differences and similarities between various scripts. Twenty-one participants were trained in crime scripting, and tasked to produce individual scripts based on the same video footage of a shop robbery. Content analysis was applied to the scripts, which involved classifying the different steps of the crime commission process and analyzing their distributions. A scoring system was then developed to assess the relative degree of completeness of each script, and linear regression computed using the number of activities included as the predictor variable. This research provides the first evidence of the limits of creating scripts using an intuitive approach, and the need for applying semistructured goal-based methods. 1