Measuring the acoustic performance of open-plan offices is an increasingly important issue for companies and employees. Indeed, it has been highlighted in several studies that noise in such workplaces reduces employees' performance and makes them be unsatisfied with their work environment. This is especially important for conversational noise. The ability of a room and its layout to reduce conversational noise can be assessed using acoustic indicators and in particular the D2S (spatial decay of the A-weighted level of a speech signal when doubling the distance from the source on a line passing over workstations). This indicator is now widely used in the field because it is recommended by several national standards that give values to be achieved. However, to date, there are no studies that discuss the accuracy of a measurement in an office. The purpose of this presentation is to provide an analysis of the measurement uncertainties for this indicator. To that end, the measurement principle of D2S will be presented as well as an analysis of uncertainties based on their theoretical developments. This analysis will be conducted using real cases (measurements) and simulated cases using a room acoustics software.