Being ‘scholarly’ includes the pursuit of grants, which requires understanding and satisfying the review criteria of specific funding organizations. An important merit review criterion against which the National Science Foundation (NSF) evaluates grant proposals is Broader Impacts (BI). The two-fold purpose of this study was to 1) identify the rhetorical conventions of stand-alone BI sections, which are expected to demonstrate the potential of a proposed project to benefit society, and 2) compare the use of rhetorical conventions in the BI sections of funded and non-funded proposals. In the tradition of genre theory, the study employed a top-down move analysis of a corpus of 91 BI texts from proposals in different disciplines submitted to the NSF. The analysis yielded a descriptive model of 3 moves and 9 steps, named Contextualize-Demonstrate-Predict, which was applied to the annotation of the entire corpus. Descriptive and statistical analyses of the annotated data provided a rich description of the composition of BI discourse in terms of primary and secondary rhetorical functions, also revealing similarities and differences in move and step distribution, functional prominence, and language use in the BIs of funded and non-funded proposals. The results of this study lend themselves to practical implications for grant writer education in rhetorical competence.