International Organizations and Machine Translation (MT) share a long history together; however, there is still some resistance to the introduction of MT tools into the institutional translation process. Additionally, organizations might not be able to select the tools that suit them better. Currently, a number of customizable statistical systems are being offered as additional tools for language support. This study describes the design and execution of a comparative context-oriented evaluation of statistical engines in the context of international organizations, with the purpose of proposing a short practical framework for MT quality assessment, corroborating assumptions about customizable and generic statistical systems, and stressing the links between MT and institutional translation.