In this paper we seek to realise the potential that Spaapen and van Drooge’s productive interactions concept offers, but which we argue has been lost through its operationalisation as a process of ‘counting interactions’. Productive interactions arise through moments of contact between two very different systems (the societal and the scientific), and each system values societal impact in very different ways. Finding mutual value in that interaction is important, and we argue that value in both arises when network arrangements shift, as academic disciplines solve urgent scientific problems and as societies improve living conditions. Productive interactions approach assumes the value-frameworks of the wider networks within which particular knowledge sets become actionable. However, our constructive critique highlights the omission of the wider elements of science and social systems within which productive interactions takes place (and whose dynamics ultimately determine the final scientific and societal impact of that research). Indeed, research evaluation to date has not considered the consequences of the productive interactions in terms of these changing relationships. To contribute to this lacuna, we propose a model that conceptualises a meso-level system comprising interactions between actors within two subsystems, highlighting the importance of coupling between researchers and users, valuation signals given to particular productive interactions from researcher and societal communities and the way these signals in turn embed useful knowledge practices. We apply it to a set of examples of productive interactions in the field of social sciences and humanities (SSH) gathered in the framework of a European project.