Rollout of development interventions using a one-size-fits-all model can achieve economies of scale but neglects to account for variability in farm and farmer characteristics. A data-driven approach to incorporate farmer diversity in scaling strategies may help to achieve greater development impact. However, interpreting the multiplicity of smallholder characteristics is complex, time-consuming, and the ways in which the insights gained can be implemented is poorly understood. Navigating these tensions, we present a farm typology study carried out in collaboration with a large development organisation (the “scaling partner”) promoting agricultural inputs in Rwanda. This study was conducted late in the scaling pathway, in order to finesse the scaling strategy, rather than to target intervention selection. Drawing on nearly 3000 interviews from 17 districts of the Western, Southern, and Eastern Provinces of Rwanda, the typology differentiates households along two axes: 1. prosperity (a cornerstone of conventional typologies), and 2. adoption of inputs (fertilisers and improved crop varieties). We used an efficient household survey tool, a minimum-variable approach, and concepts from the study of adoption of agricultural innovations. Through an action-research collaboration with the scaling organisation we adapted the methods and the findings to be “actionable. Approximately two-thirds of the study population were using fertilisers and improved seed to some extent. Along each prosperity stratum, however, there were multiple degrees of adoption, demonstrating the value of including adoption information in typology constructions. Ten farm types were identified, where the key differences along the prosperity axis were land area cultivated and livestock owned, and the key differences along the adoption axis were perceptions of input efficacy, access to training, and education level. We also present a simple decision tree model to assign new households to a farm type. The findings were used in three ways by the scaling organisation: (i) characterisation of the population into discrete groups; (ii) prioritisation, of farm types for engagement, and geographical locations for further investment; and (iii) design of decision support tools or re-design of packages to support technology adoption for specific farm types. The need for field-level validation of the typologies was also stressed by the scaling organisation.