Widespread discourses of evidence use in policy claim that by grounding development policies in evidence more effective and cost-efficient interventions will be implemented. While many decision-makers and researchers present evidence as a technical process for identifying “what works,, others argue that policy-making is run through with politics and power. We subject these claims to bibliometric analyses, mapping the structure of academic research communities for evidence use in public health and education research. Despite many differences in the structuring of these two intellectual communities, the similarities between them reveal how politics and power intersect with evidence production and use, and also affect the evidence-informed recommendations provided. Our analyses thus undermine claims of rationality and context-free “what works” agendas in evidence use in development, revealing that researchers overwhelmingly fail to cite works from the very geographies and epistemologies that are most affected by evidence-informed recommendations.