Take a loaded round, measure the neck with a bullet seated, take a sized case with no bullet, measure it, the second measurement should be .002"-.004" LESS than the first measurement.
If it isn't, take the first measurement, subtract the amount of tension you want, say a loaded round measures .364", then you want .002" tension, you order a .362" bushing.
The standard is to subtract 2thou from loaded (regardless of what you want), because necks only spring back ~1thou max, and this is what grips bullet bearing. So this is how it works: You size a neck down 2thou under cal, when the bushing is pulled the neck springs back outward ~1thou, leaving ~1thou interference fit. Then you seat a bullet, expanding(not sizing) that neck the remaining ~1thou, which represent maximum grip (full spring back against bullet bearing).
Expecting 2-4thou interference fit is only excessive. This neck condition will just be sized back up when the bullet is seated, and bullets should not be used as expanders. This still provides no more than ~1thou of spring back against a bullet. You could set it 10thou under cal, and all this does is create even greater seating force, messing with resultant CBTO, and greatly work hardening the brass. Don't believe me? Size a neck down no more than seated bearing length, any amount you want. Seat a bullet, measure loaded neck, pull the bullet, measure the empty neck. It will measure ~1thou smaller than loaded (spring back),, that's it. And that's why the recommended 2thou under loaded diameter for bushings. Any more is wrong.
There are adjustments to this though.
If you have big neck clearances the fired necks will be left a long way from desired 1thou under after spring back. If greater than 5thou, the bushing will downsize a neck more than stamped on it. This because of high sizing angle, rolling neck brass smaller than desired/predicted. This brings in trial & error for bushing selection. Expect it if you're running high neck clearances, and accept that you'll be working the brass a great amount. That's not free.
Also, we have adjustment of tension via length of neck sizing. Apply that ~1thou of spring back against another 1/16" of bullet bearing and bullet grip goes up that much. Where you actually need far greater neck tension, you could size a length of neck beyond seated bullet bearing (like with FL sizing). With this, there is the set interference fit that is not being sized back up by a seated bullet, and this area is then seriously pulling inward on the back of the bullet bearing. Normally this is donut area, which varies in thickness, causing great variance in tension provided. In fact, any time tension is increased, tension variance increases with it.
Since tension variance is very bad for hunting capacity accuracy, you should never ever FL size necks there.
For larger cartridges, running slower powders, you actually want as little tension as you can get away with, and for most people(by far) the normal ~1thou spring back against 1/8-1/4" of seated bullet bearing is plenty enough.
There are competitive underbore cartridges that rely on extreme pressure nodes. This is a huge work-around for many things, one of them being tension variance. So they can FL size necks, to cause an extreme starting pressure, with no penalty from tension variance in it. Some competitors do this, some jam bullets to do it, some both. But it's not viable in larger area cartridges, as they cannot handle very high pressures like this.
