To measure true runout, the case necks must first be turned to uniform thickness, heat treated, full length sized without the decapping rod assembly, then neck sized on a mandrel. This will mostly eliminate runout, irrespective of which full length sizing die or reloading press is used.

If you have not turned your necks to uniformity, part of the runout number is the variation in neck thickness.

If you have not heat treated the brass, part of the runout number is the deformation due to uneven stress buildup in the neck of the brass.

There are countless videos posted on how changing a die brand or press brand has improved the measured runout. That in itself is not able to consistently improve the runout. A sizing die with no decapping rod assembly does not induce runout. A press does not induce runout. Something else was done, changed, or used to affect the outcome at the same time the equipment was changed, giving the impression that the equipment change was the sole cause.

In all these instances, the measured runout will increase again over the lifetime of the brass if the the necks are not turned, heat treated and mandrel sized.

If changing the sizing die and using it with the decapping rod assembly improves runout for you, it means your new die has better alignment of the decapping rod assembly than the old die. Over time, this alignment can shift, which is why if you want better concentricity, just remove the decapping rod assembly instead.

If changing the press only has improved runout for you, most likely the brass used is not from the same batch, does not have the same amount of firings, or was prepped differently. This can be demonstrated by mounting the old press and new press on the bench next to each other. Use the same sizing die on both presses. Pick out brass of the same headstamp, with the same amount of firings of the same load, and from the same batch that have been prepped the same. Size a few in the one press, and a few in the other. Measure the runout of the loaded rounds. Is there a difference ?

The seating stem is very seldom the cause of runout, but when it is, it is very difficult to debug.

Any work done towards making reloads more consistent can improve accuracy. There is a point beyond which additional work is pointless.

A lot of time better spent on the trigger is wasted trying to obtain certain measured parameters, especially when what you are actually measuring is not what you think you are measuring.