This study contributes to the ongoing discussion on how to attribute and evaluate the contribution of transdisciplinary research to sustainable development. As co-created knowledge is a key product of transdisciplinary research, we tested the hypothesis that the extent to which this knowledge is utilized beyond the project consortia, in different areas - from scientific methods and insights to policy decisions - and across a continuum of geographical scales can be used to identify potential impact pathways. For this purpose, we developed an analytical framework that links the transdisciplinary process to six possible utilization stages, which we used as indicators of the usability of co-created knowledge. We gathered data from 22 research projects active in 36 countries using a survey and semi-structured interviews. Our results show that even during implementation of the projects, co-created knowledge is utilized by multiple actors at different stages, in all areas and at all scales simultaneously, suggesting multiple impact pathways. Project knowledge is predominantly utilized for national-level policymaking, and research partners named co-creation of knowledge with key stakeholders as the most frequently used mechanism for promoting knowledge utilization. Closer analysis revealed different understandings of and approaches to knowledge co-creation. These can be linked to weaker or stronger definitions of transdisciplinarity. The analysis shows that researchers using strong transdisciplinarity approaches typically face challenges in encompassing multiple epistemologies and facilitating dialogue. Some results suggest that inclusion and collaboration in co-creating knowledge can empower actors otherwise excluded. Our research shows that although transdisciplinary projects have nonlinear impact pathways, these can be partially assessed using the proposed analytical framework. Further, our results indicate a link between usability, inclusion, and collaboration in transdisciplinary research. We conclude with the observation that transdisciplinarity and its requirements still need to be better understood by actors within and beyond the research community.