It's simple combinatorics.
You do a 10 shot doubles strings (5 doubles) with each combination of possible changes in your grip, trigger finger placement, and engagement. Since you can't tell ahead of time which of these variables can be safely merged as a single contributing variable, you need to test them all independently of one another, which leads to a multiplicative combination of all possibilities in order to actually get enough data to do a factor analysis to identify what actually does and does not contribute the highest statistically significant predictive power towards "good shooting."
Assuming you can come up with a time and score based metric that you're happy with to evaluate each combination, then you simply vary each variable independently of the others until you have tested them all, trying to hold energy and mental focus constant to avoid introducing those as a variable.
So, some variables would be:
- Degree of grip strength (4 possibilities)
- Amount of wrist lock on the pinky side (On or Off, 2 options)
- Grip position of the hands (4)
- Support hand placement (2 or 3)
- Sideways pressure on the gun (2 or 3)
- Clamshell pressure or not (2)
- Elbows up or down
- Push pull light or heavy (2)
- Horizontal isometric tension or not (2)
- Bicep and Tricep pre-engagement to lock the elbows (2)
- Locking the elbows out or bent (2)
- Degree of lean of the body (2 or 3)
- Feet position (2)
- Engagement of the shoulder girdle and upper body to lock the shoulders and/or apply tension in the upper body during firing (2)
- Trigger finger placement (3)
- Degree of trigger pin (2)
- Snatch or Slack trigger approaches (2)
- Degree of shoulder elevation (2)
- Head angle (2)
- Degree of finger engagement up and down the hands (3x3)
- Degree of muscular engagement in the core (2)
- Degree of engagement with the thumbs (2)
It's nothing more than basic scientific and statistical practice and high school combinatorics.
Obviously people don't do this. Instead, they try things based on intuitive triggers that play with a number of these variables at once and then figure out what appears to work for them and what doesn't based on rough timings. I doubt that many people in their timings have done statistical timings to determine the significance of the change they made and determine the variance in their behaviors in order to tell whether a change was the result of something that they think was causing it or whether it might have been something else.
But that's the nature of human performance. It is exceptionally difficult to quantify and analyze. Moreover, the people who are good at performing often aren't sure why they are doing something, which makes it hard.
ETA: the above list is just a simple off the top of my head collection of the various different elements of shooting that various top trainers I have watched, read, or observed have said or mentioned in regard to the gripping and recoil control process.