Impact assessments are used to raise evidence and guide the implementation of sustainability strategies in commodity value chains. Due to methodological and data difficulties, most assessments of agricultural commodities capture the impacts occurring at the farm-level but often dismiss or oversimplify the impacts caused by land use dynamics at larger geographic scale. In this study we analyzed the impacts of two cocoa production systems, full-sun and agroforestry, at the farm-level and beyond the farm-level. We used life cycle assessment to calculate the impacts at the farm-level and a combination of land use modelling with spatial analysis to calculate the impacts beyond the farm-level. We applied this to three different future cocoa production scenarios. The impacts at the farm-level showed that, due to lower yields, cocoa agroforestry performs worse than cocoa full-sun for most impact indicators. However, the impacts beyond the farm-level showed that promoting cocoa agroforestry in the landscape can bring the largest gains in carbon and biodiversity. A scenario analysis of the impacts at the landscape-level showed large nuances depending on the cocoa farming system adopted, market dynamics, and nature conservation policies. The analysis indicated that increasing cocoa demand does not necessarily result in negative impacts for carbon stocks and biodiversity, if sustainable land management and sustainable intensification are adopted. Landscape-level impacts can be larger than farm-level impacts or show completely opposite direction, which highlights the need to complement farm-level assessments with assessments accounting for land use dynamics beyond the farm-level.