Ram extrusion of a solid granular soap was studied using three geometrically identical but differently-scaled extruders. The experimental design revealed deviation from the Benbow and Bridgwater (1993) extrusion model due to nonideal, scale-dependent effects. Typically these effects, linked to the shear rate in the extruder, are absorbed into the model's material pseudo-properties. The data were able to be represented using the Basterfield et al. (2005) model for extrusion flow which does include a shear rate as a variable. Flow visualisation in conjunction with fluid dynamics-based simulations showed, however, that the assumptions underlying the Basterfield et al. model are not appropriate for soap extrusion, despite the good agreement of the model with the experimental extrusion data. This highlights a need for care in interpretation of extrusion data, in that the limited information gathered about any given experiment, typically just the extrusion pressure, can lead to the generation of spurious parameters if the wrong model is applied.