Accounting for similarity among alternatives is important for having unbiased estimates and behaviourally reasonable substitutions. Capturing similarity in a spatial context is a challenging task and the common approach of discretising space into a number of disjoint nests generally leads to uncaptured spatial correlations. On the other hand, relying on more complex error structures quickly leads to computational issues. In the present paper, we propose an alternative approach, where a Cross-Nested Logit (CNL) modelling framework with a flexible correlation structure is used, where space is treated as continuous, while the allocation can be parameterised based on a range of similarity factors. The proposed structure is applied in the context of mode and destination choices of shopping trips using a smart-phone GPS panel survey from Leeds, UK. Results indicate that in addition to the improvements in model fit, the proposed CNL specification is able to uncover interesting findings about individual mobility behaviour.