In this study, we investigate the feasibility of using a vibration-based metric for predicting low-frequency structure-borne noise in cross-laminated timber (CLT) buildings. The overall aim here is to facilitate the conceptual design for CLT buildings in terms of estimating noise levels. Since noise levels are inherently sensitive to architectural or mechanical design changes, in the low frequency regime, the problem of predicting these as a metric is well-known as a “high in computational cost” but “low in confidence solution” problem. Here, a reduced and robust prediction metric offers a way forward in estimating intrusive noise levels while capturing the effect of conceptual design changes. The degree of correlation between the vibration-based metric and the predicted noise levels was investigated by performing linear regression on datasets generated from a parameterised finite element model of a CLT building structure. A generally high degree of correlation between the vibration-based metric and the noise levels is observed. The effect of employing various frequency bands and sets of vibration evaluation points are investigated, and it is concluded that a high degree of correlation is obtained using only 3 × 3 vibration evaluation points per CLT panel.