In the design process of extrusion dies, the utilisation of computational fluid dynamics (CFD) is state of the art. To evaluate the distribution quality of different geometries, the circumferential velocity at die exit is the most commonly used criterion. In conjunction with maximal and minimal wall shear stresses, it is possible to define an overall quality score for each die based on CFD results. In the face of shrinking batch sizes and the requirement of more frequent changes of colour or material, the localisation of stagnation zones becomes more and more important. These stagnation zones lead to higher residence times of the polymer melt inside the die. Thus, it takes longer to clean the die and it subsequently requires more purging material. Therefore, new quality criteria for extrusion dies should be taken into consideration to detect regions with high residence times. In post processing, streamlines could be used to track down stagnation zones qualitatively. However, this method involves user interaction as well as additional computations. To obtain the spatial distribution of residence times from streamlines, further calculations are required, making this method even less suitable for automatic geometry optimisations. Instead, it is proposed to use the concept of the local mean age of melt, which is directly obtained from the numerical solver calculating an additional transport equation. The age of the melt is available as a scalar field, making it simple to estimate e.g. global extreme values. Based on the mean age, new quality criteria are developed for the automatic optimisation of spiral mandrel dies. This approach ensures that problematic spots such as the feed-in region of the spiral grooves could be safely detected and locally optimised.