This article is based on our review and analysis of French texts from a corpus of gallery signs and artwork labels used in international painting exhibitions. Anomalies identified in these museum texts (source and target) were sorted out according to three different criteria: their nature, their effects on the visitor/reader, and their origins (anomalies present in the original text or anomalies due to a translation error). By looking into the correlation between correctness, consistency and efficiency (or lack thereof) of a given section from the source and target material, we can gain insight into the weaknesses-to be remedied-but also the strengths to be promoted-of the process by which those involved in collaboratively creating the material develop its editorial quality. Our analysis is based on examples that illustrate the three cases we have modelled: (1) anomalies identified in the original text and transferred into the French translation; (2) lower editorial quality of the French translation compared to the original text; (3) higher editorial quality of the translation compared to the original French text. The examples are then analysed through the lens of emerging good practices for editorial quality assurance in the multilingual publishing industry. These combined good practices, adapted to the characteristics specific to informational texts used for exhibitions, form the basis of our recommendations intended for those in charge of multilingual communication in museums looking to improve risk management in multilingual publishing and to take advantage of synergies resulting from an editorial process involving three people (or more).