Okay, I specifically opened the game for a few minutes just to test how the Blunderbuss (no reason to believe GR is any different) headshotting works. I used Huntsman, mostly because he has an innate special to indicate effective headshots, but I happened to have a Conservative BB too.
The result was inconsistent at best. I guess high-quality slo-mo video could help more, but oh, well. The only consistent thing I could get out was that the bash either cannot headshot (probably due to being a genuine AoE effect) or is extremely hard to do so with. So that’s no use. Otherwise, my best deduction is that of the pellets that hit, the one that’s closest to the center of the aiming reticule decides whether a “real” headshot occurs. Only when I aimed at a dummy’s head, next to it or otherwise near it did ammo recovery occur. The visual indicator of a headshot - a different-colored crosshair - seems to be reliable indicator in this case, as every time it indicated a headshot, I gained ammo. I also got several times when the audio indicator of a headshot played, but only damage was affected, so those seem to be interlinked. Those times I was aiming at the torso of the dummy, or at torso height from further away, and at that point neither the visual indicator nor the ammo recovery worked.
If my testing is correct, I’m not sure what to make of it. On one hand, it indicates that to use Conservative Shooter, Huntsman’s passive, or other possible headshot abilities, you still need skill, but it mitigates the advantage of a shotgun-style pellet pattern to begin with, as your aim needs to be pretty good to reliably get the ability-proccing headshots. On longer ranges, it also gets to be a matter of luck. The procs are already limited to one per shot; every time I gained ammo I gained two (accounting for the passive and Conservative both), no matter how many pellets hit, so that’s not why its constructed like this.
I’d like to have each pellet check for headshot (not just damage) separately, but the procs still limited to one per shot, if only to play around and test how well it would work (primarily on Huntsman).
Note that this was a test done by one person in a few minutes and in the Keep, so it doesn’t reflect how well this works in-game at all, and isn’t sufficient to make assumptions with significant certainty.