The field of knowledge brokering in education--aiming to better connect research to practice--is currently emerging. Evidence of a community dissonance between researchers and practitioners in education suggests that models of knowledge brokering that consider the perspectives and priorities of both groups are required. It is also a priority to identify what kinds of knowledge brokering products such as research summaries are successful at communicating research to teachers, potentially functioning as boundary objects. We report findings from a comparative judgement study where a group of 28 mathematics teachers and a group of 19 mathematics education researchers ranked twenty research summaries, from different sources, in terms of their success at communicating the research to teachers, and explained which features contributed to this ranking. Overall the findings suggest moderate consensus both within and between the two groups, and some important areas of difference. Both groups agreed that graphic design was the most important element of a research summary; that being summarised, having implications for practice and being easy to read or accessible and well-structured were key features of a research summary; and that the length of the research summary or the time it might take to read were also important. Whereas teachers mentioned the choice of topic of the research summary and to some extent language more than researchers, researchers highlighted some other features of successful research summaries for them that teachers did not: ideas around methodology and use of evidence, opportunities for critical reflection, and issues of trust and credibility.