First, pick your test parameters and stick with them forever.
I found that there were differences with SD based on 5 round strings vs 10 round strings, but not significantly so between 10 round strings and 20 round strings, so I do 10 round strings.
After you have a lot of test data in the books, you can take a bullet you have never shot, and you can reliably predict how it will perform with a powder you are familiar with.
Like Painter, I will break a ladder up based on the size of the load window. If my load window is 3.6 - 4.2, my ladder probably goes:
3.6
3.8
4.0
4.1
4.2
I used to test accuracy (and chrono) at 30 feet. When I switched to coated lead, I found that groups were simply TOO similar, so I switched to 45 feet to make differences more easily visible and measured. Sticking with same distance is valuable as it allows you to look at a lot of targets and simply by visully comparing them to past targets in memory, say "Nope, not acceptable. This does not meet my standard."
After identifying most accurate load, retest at/around that load with small OAL tweaks.
When you chrono or test accuracy, tap the magazine back down on the tabletop before inserting into pistol. This breaks up any clumps, improves consistency of powder position, and makes the burn more consistent. I can't say for sure that it improves accuracy, but it does improve SD.
edited to fix a couple of type-os and improve clarity