Urban Mobility Decision Support Systems help policy makers in designing the actions that will improve urban mobility. Typically, this is done via policy optimisation techniques. Current approaches to policy optimisation are based on iterative simulations of a model in order to find an optimal policy combination. Such approaches require fast simulation procedures, as the simulations are performed repetitively with varying parameters. Simulations at a micro level present higher accuracy but require longer execution times, and therefore they can not be applied to current approaches without resulting in slow optimisation operations. In this paper we present an approach for policy optimisation based on micro-level simulations where an heuristic provided by the policy maker guides the selection of the scenarios to be simulated. This allows to set a limit on a potentially big search space, and allows for a more accurate selection of the simulation scenarios.