You can minimize some variation by shooting "round robin" style. Yesterday I was testing 6 different seating depths. I loaded 60 identical rounds, differing only in seating depths, ten rounds at each depth I was interested in testing.
I printed an 8 1/2" x 11" card with six small targets, one target for each of the 6 different seating depths. I shot them in sequence, one round at a time for each of the different seating depths. This requires that you carefully position the correct round in your ammo box when you're loading and label them properly so you can keep track of what's what. You need to pay attention at the range too.
I repeated cycling through the six targets until each one had five hits. In other words, rather than shoot five identical rounds in a row at target one, I shot only one round, moved to the next target and shot a round which was seated slightly longer, and so on, making sure that as I fired each round, I launched it at the correct target.
That way I ended up with a five round group on each target, but spread out over time to equalize things like wind, barrel temperature, fouling, etc. Actually, I wasn't equalizing so much as spreading possible errors evenly among the 6 test subjects, but you get the idea.
Then I switched to card #2 and did the same thing for the remaining 30 rounds. I went home with two cards, each with five shot groups for each of the 6 variables I was testing. The variations were small, but after scanning the targets, doing some careful computer scoring, and then analyzing the data in M.S. Excel (along with making several graphs), I was able to detect a clear winner.
As you might suspect, during my test the conditions changed from dead calm to a light breeze, the barrel warmed up and cooled down a number of times, and the degree of fouling changed too. Had I not used this round robin style of testing, I wouldn't have as much confidence that I properly extracted the test results I was looking for.