With Web-based e-services called "election engines," citizens can search for and compare candidates in local and national elections using issue-based questionnaires. The results of such services are generally in the form of search engine lists. However, data on political issues and agendas is very complex, i.e., highly multidimensional. Hence, we argue that relevant voting decisions should not be based on simple search engine results, but on more encompassing cognitive operations. In our project, the service was redesigned to extend the function from a mere search engine to a navigation toolset with real-time feedback. Two types of visualization components, a similarity map and sector diagrams, help citizens to make sense of multidimensional political spaces by facilitating exploration at will. As the result of an extended research project, our redesigned service was published by a major commercial broadcasting company for the Finnish EU elections in June 2004 and in local elections in October 2004.