Abstract Reverberation mapping (RM) is an important technique in studies of active galactic nuclei (AGN). The key idea of RM is to measure the time lag τ between variations in the continuum emission from the accretion disc and subsequent response of the broad-line region (BLR). The measurement of τ is typically used to estimate the physical size of the BLR and is combined with other measurements to estimate the black hole mass MBH. A major difficulty with RM campaigns is the large amount of data needed to measure τ. Recently, Fine et al. introduced a new approach to RM where the BLR light curve is sparsely sampled, but this is counteracted by observing a large sample of AGN, rather than a single system. The results are combined to infer properties of the sample of AGN. In this Letter, we implement this method using a hierarchical Bayesian model and contrast this with the results from the previous stacked cross-correlation technique. We find that our inferences are more precise and allow for more straightforward interpretation than the stacked cross-correlation results.