An effective decision support system requires a user’s trust in its results, which are based on expected utilities of different action plans. As such, a result needs to be explainable and explorable, providing alternatives and additional information in a proactive way, instead of retroactively answering follow-up questions to a single action plan as output. Therefore, this paper presents LEEDS, an algorithm that computes alternative action plans, identifies groups of interest, and answers marginal queries for those groups to provide a comprehensive overview supporting a user. LEEDS leverages the strengths of gate models, lifting, and the switched lifted junction tree algorithm for efficient explainable and explorable decision support.