I always liked to shoot five shot groups because that is how I came up doing it. You must include all groups and all shots though. I was used to to concentrating for five shots, taking a break, shoot five more etc, but I have in the last three or four years shot some reduced target F Class where we shot 20 rounds each target. Either way, you get a big sample. A 20 shot group, or a 20 or 25 shot "aggregate". Either way it counts all shots. And of course it includes shooter error, both physical error and wind/mirage reading error. (Some days I just suck and pack up and go home.)
Ladder schmatter, OCW or whatever, we have to put shots on target and see what it does. That's all I have ever done to find a good load. Try things and shoot them a lot. Early on in testing (Not many shots) you may be encouraged, and it may hold up (Although it doesn't always), but you have to keep shooting and see.
And I have said for years here a good pistol load will shoot over a small spread. If 4.0 shots great but 3.9 or 4.1 doesn't, 4.0 was probably a fluke. It's no different for rifle, no great loads only shoot at "X" powder weight but not .1 up or down.
And with those great loads we will shoot the occasional screamer group, but then miss a condition and shoot a big one, or the law of averages catches up with us and despite our best effort we shoot a mediocre one.
But we see it all the time here where someone shoots one group with each charge weight and then asks which one is best. One group is meaningless. It may be worth pursuing, but it means nothing by its self.