Environmental impact objectives are commonly found in building performance labels and rating schemes. Anticipating a building’s impact from the conceptual design stage and identifying decisions that do not compromise its chances of reaching these targets is therefore crucial. Yet, few methods and tools are able to provide tangible decision support through a context-specific and early-stage-oriented approach. This paper proposes a workflow to do so based on a generative approach and interactive decision trees. Illustrated on a case study, the approach consists in generating building scenarios by varying parameters not yet fixed at the early stage, including geometrical (e.g. building shape and height), architectural (e.g. façade opening ratio) and technical (e.g. heating system) parameters. The series of scenarios are evaluated in terms of their greenhouse gas (GHG) emissions over their life cycle (including construction and operation), as well as from building-induced mobility. The effects of filtering this database according to a given impact target are explored using a classification algorithm that produces a decision tree showing the proportion of target-complying and non-complying scenarios, as well as the (un)favourable decision pathways. Stakeholders of the planning and design process can therefore get insights into the implications of a given string of decisions.