We extend teacher evaluation research by estimating a reformed evaluation system's plausibly causal average effects on rural student achievement, identifying the settings where evaluation works, and incorporating evaluation expenditures. That the literature omits these contributions is concerning as research implies it hinders evidence-based teacher evaluation policymaking for rural districts, which outnumber urban districts. We apply a difference-in-differences framework to Missouri administrative data. Missouri districts could design and maintain reformed systems or outsource these tasks for a small fee to organizations like the Network for Educator Effectiveness (NEE), an evaluation system created for rural users. NEE does not affect student achievement on average but it improves math, and possibly reading, achievement in rural schools where the average student's prior-year achievement score is below the state average or the average teacher's years of experience are below the state average.