In recent years, the concept of Business Intelligence (BI) has become increasingly popular in both academic and practitioner literatures. Many universities have been adopting BI systems to help them understand, improve, and report on their performance, a trend that is accelerating with the rise of influential and controversial assessment regimes. The core assumption of BI systems is that they offer a ‘single-source of truth,’ but such an assumption may not be tenable in an academic setting because much academic research is a social construction and academics and administrators can shape how they present their research to others. Motivated by the practical importance of effective use in this context, we conducted an in-depth single case study to learn what it would take for an organization such as a university to use a BI system effectively. We focused particularly how the concept of representational fidelity and how the university represented its performance in a national assessment exercise. Our data analysis revealed the practices involved in the production of representational fidelity at the individual and collective levels, and the interdependencies between these levels. From the findings, we developed a theoretical framework to explain how individuals participate in the production of representational fidelity. More generally, we introduce a new perspective on the role of BI systems, i.e. BI systems are used to optimise objective data and serve as a rose-tinted window. We also discovered conditions that enable and constrain this process.