And by simple, you only need three parameters: the mean branch length of a phylogeny in subs/site, the number of sequences in your tree, and an estimate of the per-generation evolutionary rate. The formula seems to work surprisingly well on simulated outbreak datasets given its simplicity!
