Tuning a load and then maintaining the tune is something all reloaders go through, and there are a variety of different methods that can be used, largely dictated by the precision requirements and/or specific shooting discipline or use for the load. For many applications, one could use different approaches and still end up at roughly the same place with regard to precision, although you will find that those striving for the very utmost in precision often go far beyond what the typical recreational shooter might do. In this response, I am going try to avoid going into a level of detail that might be applicable for someone striving for uber-precision, and keep things to a more basic level.
One example of how different approaches can lead one to essentially the same place in terms of basic reloading practices would be using a ladder test versus an optimal charge weight test. The ladder test, attributed to Creighton Audette (i.e. do a search for Audette Ladder Test), is set up such that single rounds loaded at increasing charge weights are fired at some distance, usually at least 300-400 yd or preferably even greater, using the same point of aim for each shot. The idea is that as charge weight increases, velocity will generally increase, and ultimately generate a [somewhat] vertical string of impacts on the target. The main point of the Audette ladder test is to look for at least two or three successive charge weights where the bullets impacted the target at very close to the same vertical height above the point of aim. The idea behind this is mainly about barrel harmonics and the launch angle of the barrel out toward the muzzle when the bullets actually exit the bore. Within a certain charge weight test window, bullets having a slightly faster velocity exit the bore as it is on the upswing in the harmonic cycle. Bullets having a slightly slower velocity will exit the bore slightly later during the cycle, when the muzzle is pointing upward at a slightly higher angle. Because the slightly slower bullets exited the bore when it had a slightly higher launch angle, they impact the target at approximately the same vertical height as the slightly faster bullets that left when the barrel launch angle was not as high. So even though they had a bit less velocity, the slower bullets were launched at a slightly higher angle and therefore hit the target at approximately the same vertical height. This effect is known as positive compensation, and probably works best over a velocity spread of about 20-25 fps. For that reason, what you'd expect to see on a ladder test is a series of impacts moving up the target as charge weight increased, but where some of the successive charge weights gave impacts that were grouped close together vertically, whereas others might be much farther apart with each successive charge weight increment. What you're really looking for is those successive charge weights (i.e. a charge weight window) where the bullets grouped close together vertically. These charge weight windows are often referred to as "nodes".
The Optimal Charge Weight Test is attributed to Dan Newberry (i.e. do a search for Newberry OCW Test), and uses groups rather than single shots. The key to the OCW test is finding a few successive charge weights where the center-point of each group is positioned the same with respect to the point of aim. To be clear, you are not looking for the tightest group, you are looking for two or three successive charge weights where the group center-points are not moving around with respect to the point of aim. The underlying premise of this test is exactly the same as the ladder test. That is, finding a charge weight window where slight changes in muzzle velocity are accompanied by corresponding changes in barrel harmonics that minimize the effect of velocity variance such that the shots largely impact the same point on the target (i.e. better precision). Although it can certainly be done at longer ranges, it is common to carry out an OCW test at 100 yd.
There are advantages and disadvantages to each approach. For example, the OCW approach may be thought of as less sensitive to wind conditions because of the much shorter distances over which it is usually carried out. However, it is also somewhat less sensitive for the same reason. Because it is fired as single shots rather than groups, the ladder test typically can be carried out with fewer loaded rounds, even if you carry it out in duplicate, and it can be very sensitive if carried out at distances of at least 500-600 yd. The downside to that is that one must have access to a range of sufficient length in order to effectively carry out a ladder test, as trying to do a ladder test at only 100 yd makes it far too difficult to reliably interpret the pattern of impacts.
So how do these test fit into a typical load development test process? A common approach is to first find the optimal charge weight window, then tune groups using seating depth. Although that is probably the order most commonly used, one could certainly do a coarse seating depth test initially to find some seating depth the bullet seems to like, then do the charge weight testing and follow up with a fine-increment seating depth test. The caveat to that approach would be that the charge weight used for the coarse seating depth test would likely have to be an educated guess, at best, as the optimal charge weight would not yet have been identified via testing. Nonetheless, it can be done that way and still end up at more or less the same spot. The closer the charge weight used for an initial coarse seating depth test is to that of a final tuned load, the better. Unfortunately, we don't always know in advance where a given load will tune in in terms of charge weight.
I have used the term "windows" several times in this response. It is important to note that finding an optimal window means the test data have to define both edges of the window, low and high. Otherwise, the window is not really well-defined and there is no way to know with any certainty exactly where the middle of the window actually resides. The whole point of identifying an optimal window is so you can load to the center of it, thereby gaining the most "forgiveness" with the load, and increasing the odds the load will continue to shoot well for some time before it eventually goes out of tune. This is a concept you may have already run up against because it is not uncommon where someone has done random spot testing and somehow managed to come up with a fairly decent load, only to find that it goes out of tune rather quickly, or even vanishes entirely when the conditions change. If some parameter of a rather randomly-identified load such as charge weight just barely manages to catch one edge of an optimal window, it may shoot well on that particular day, or even for a little while. But because that parameter was at the edge rather than in the center of the window, it takes very little change for the load to go out of tune. If you get nothing else useful out of this post, I hope you will appreciate the value of defining both sides/edges of optimal windows through rigorous testing across a sufficiently wide range in fine increments. There is really no other way to do proper load development, because anything else is really a "guess" or just "luck".
