The ISO 354:2003 standard relating to sound absorption measurements is currently under revision to improve the reproducibility of the procedure it describes. Round robin tests conducted across various reverberation rooms indeed revealed significant disparities between sound absorption measurements of the same sample. One of the reasons is that, at low frequencies, the sound field in a single laboratory cannot be considered fully diffuse. However, the average sound field across different laboratories may be considered diffuse if the interaction between the finite sample and the diffuse field is duly accounted for and the direct field close to the absorber is disregarded. In this work, a method is developed for optimizing reverberation room design such that measured absorption values are as close as possible to ensemble average diffuse values. The reverberation room is modelled using the finite element method and standardized measurements of an absorptive sample are simulated. The distance between resulting absorption coefficients and diffuse target values is minimized in an optimization procedure having the geometrical characteristics of the model as input parameters. The results are anticipated to participate to the revised ISO 354 as guidelines for the construction of new reverberation rooms or the improvement of existing ones.