Most gamma-ray burst (GRB) X-ray afterglow light curves are characterised by a plateau, followed by a normal power-law decay, which is interpreted as afterglow emission, that is radiation emitted by the shocked interstellar medium that is swept up by the blast wave. Despite the numerous alternative interpretations, the origin of the plateau remains unclear. In the early years of the Neil Gehrels Swift Observatory, it was suggested that the plateau might be afterglow radiation, that started before the prompt gamma-ray emission, and its time profile would be an artefact of assuming the start time of the prompt gamma-ray emission as zero time (the so-called "prior activity model"). We aim to test the plausibility of the prior activity model by leveraging the current Swift sample of early X-ray afterglows of GRBs with measured redshifts, which is more than eight times larger than the one originally used (463 vs. 56). We modelled the GRB rest-frame X-ray afterglow luminosities assuming a simple power-law with the true reference time preceding the prompt gamma-ray emission trigger time by T0 and the X-ray luminosity L0 at the trigger time as free parameters. We tested each case applying both chi^2^ and runs tests. For 90% GRBs of our sample, the model provided a successful description. In ten cases the afterglow peak is identified and modelled appropriately. Using the 300 GRBs with accurate parameters' estimates, we confirm the anti-correlation between L0 and T 0 with 0.7dex scatter. In addition, selecting the subsample of 180 from the literature with reliable estimates of isotropic-equivalent released energy Egamma,iso, peak luminosity Lgamma,iso, and intrinsic peak energy Ep,i of the nuFnu spectrum of the prompt gamma-ray emission, we find a correlation between L0, T0, and Egamma,iso (0.4dex scatter) over nine decades in L0 and common to all kinds of GRBs. The afterglow likely begins in most cases before the start of the detected prompt gamma-ray emission by a lognormally-distributed rest-frame delay with a mean of 10^3^s and 0.8dex dispersion. As also suggested by the recent discoveries of Einstein Probe of X-ray emission starting long before the prompt gamma-rays, our results suggest that the occurrence of prior activity could be much more frequent than what has tacitly been assumed so far.

Cone search capability for table J/A+A/703/A101/table3 (Time intervals that were excluded from the fit because of the presence of internal activity, for a restricted list of GRBs)

Cone search capability for table J/A+A/703/A101/table4 (Best-fit parameters for the main group of GRBs)

Cone search capability for table J/A+A/703/A101/table5 (Best-fit parameters for the AG-rise set)