This paper develops a renewed research impact assessment approach that goes beyond the traditional computation of internal rates of returns but allows collective learning, understanding and guiding systems transformation. Our approach (ASIRPA) builds on the broader research impact assessment literature which discusses impact-generating mechanisms in nonlinear dynamic processes. The main originality of ASIRPA is that it is based on a theory of innovation inspired by Actor Network Theory and a standardized ex-post case study methodology. Standardized case studies allow systematic codification of the variables for each case study, and the building of four ideal-type impact pathways. Each of these ideal-types is characterized by specific translation mechanisms, critical points, research and adoption networks, research outputs, and impacts. Our analytical framework and empirical analysis provide new insights into the contribution of an agricultural Public Research Organization to impact generation, and the role of users and networks in impact pathways.