At the crossroads of climate change and land competition, agrivoltaic systems (AVS) propose a land-sharing synergy between photovoltaic (PV) electricity and crop production or animal husbandry. By partially covering crops with PV panels, AVS can enhance food production by maintaining favorable microclimatic conditions and can constitute a promising adaptation to climate change. However, the AVS electric and crop production performances vary extensively across crop types, climates, panel orientations, etc. While the few existing life-cycle assessment (LCA) studies evaluating AVS’s environmental performance cover a minor part of this diversity of configurations, a planned deployment policy for AVS requires extensive knowledge on the configurations that may or may not environmentally outperform conventional alternatives, i.e., conventional PV electric and crop productions. To identify these configurations, we built a parametrized consequential LCA model that simulates an AVS driven by either electricity demand or crop demand. We then applied a scenario discovery algorithm over stochastic AVS configurations to identify groups of AVS configurations that perform better or worse than their conventional counterparts. By parametrizing the technological and agronomical contexts, the interactions between PV and crop productions, and the electricity and land markets, we provide insights regarding how AVS should be designed, promoted, or discouraged.