Optical fibre–based sensors have now established their place in the field of geohazard monitoring due to their sensitivity to strain and temperature changes. Progressive development in the technology leads to the availability of novel, accurate and durable sensors at a relatively limited cost. This creates room for original monitoring applications that have been, so far, impeded by the shortcomings of conventional monitoring tools. In this work, we explore the applicability of an interferometric optical fibre sensor as a vibration sensing tool at laboratory scale. We tested the ability of the sensor to identify precursors of instability in a downscaled model of a rainfall-induced landslide composed of granular material. We carried out four experimental tests which involved different sensor deployments and soil mixtures. The recorded signals were processed by means of a time–frequency analysis and we identified two frequency-domain parameters—the spectral centroid and band power—that could provide information on the development of instability. Their ratio yielded a unique parameter through which a precursory stage could be outlined by defining a threshold value based on the data collected at the beginning of the experiment. In our lab tests, precursors of instability were detected 2–3 min before a crack was observed at the surface. This may upscale to a lead time of about 20–30 min or more in the field, classifying our monitoring approach in between an alarm and a warning system. The work presented here can be considered a first promising step towards an innovative monitoring system and shows the potential of optical fibre sensing as a shallow landslide monitoring technique, encouraging further testing, especially in real-case studies.