Recent research on consideration set models of the vote holds important implications for the design of VAAs. Consideration set models argue that electoral decisions proceed in two stages. In a first stage, voters eliminate alternatives to derive at an individualized choice menu. They do so by relying on heuristics, i.e., relatively crude diagnostics of the appeal of different political parties. Sometimes, those heuristics suffice to narrow down the choice menu to a single party. In this case, the choice process ends. More typically, at least in multi-party systems, several alternatives will remain. In this situation, the second stage of the choice model is triggered. This stage involves a more systematic analysis of the alternatives that remain, a process that ultimately results in choosing a particular party (or in abstention). VAAs can inform both stages of the process, but require different levels of specificity at each stage. The elimination or screening stage requires relatively crude information. One can think of litmus test issues or overarching themes such as the economic role of the government, openness to new lifestyles, and acceptance of immigration. To aid voters in choosing one alternative from their personalized choice menu, far more detailed information may be needed. Consider, for example, a Swiss voter who considers both the SP and the GLP. For this voter, details on environmental plans may be decisive in how she casts her vote. Consideration set models of the vote thus require adaptive forms of advice. The present paper reviews the consideration set literature, develops proposals for adaptive VAAs, assesses the normative implications of changing VAAs in this direction, and presents results from a pilot experiment at the University of Zurich.