This article critically examines commensurability in public health – that is, the possibility of comparing phenomena according to a common standard, using indicators – as well as the scope and limits of such comparability when it is mistaken for complete understanding. It proposes a brief audit framework for constructing and interpreting indicators when commensurability is partial, using the maternal mortality ratio (MMR) as an example. With reference to the MMR, the article shows how definitions, time windows, and the quality of data recording affect comparability across contexts, and how performance pressure may induce reactive administrative or behavioral shifts, which must be empirically assessed. This paper does not argue against measurement or quantitative accountability; rather, it seeks to strengthen them through transparency of assumptions, reporting on quality and uncertainty, and the use of triangulation and auditing when incentives are high. It thus avoids both technocratic drift and the relativism that dismisses measurement on the grounds that it is constructed. It concludes by proposing that dimensions not captured by the indicator – lived experience, dignity, and trajectories of access and care – be granted specific status and formally integrated into the interpretation, through layering of evidence and case reviews.