The bushing sizes it down, that means the outside is consistent and so any discrepancies in the neck thickness gets pushed to the inside.
A mandrel will size from the inside to make the inside concentric for the surface that touches the bullet and to push the inconsistencies to the outside where they are floating in space inside you chamber.
Now, if you have a 308 with brass which necks are .013" thick then you would have .308+(.013*2)=.334" for t a loaded neck diameter.
If you were to use just a bushing only for a general starting point you would choose one that is .002" less than the loaded round diameter so .334-.002=.332" for that bushing.
If you were to want to use a mandrel that .332 bushing would be just about what you want to end up as so it would only brush against the inside, not enough displacement to actually size anything. However if you chose a .331 or .330 bushing the mandrel would then have to open the brass up .001-.002" which should be enough to push the differences in thickness to the outside and return the final result of .332".
Different thickness of neck material means different diameter bushings.
Or so the theory goes...
Whether you gain anything from it is up to you and the target.
(Edit: I have them and I use them for the stuff that Ive spent a bunch of money on brass and barrels etc because I have the time to add an extra step so I figure I might as well. For hunting and ARs I dont bother. Ive tried to test and while it does seem to maybe help my regular dies with expander balls Im hard pressed to see a difference down range. But then again I never have the wonderful tests with low single digit SD for any meaningful sample size greater than 3 without deleting the outliers. So if you want to try them by all means go for it, if youre looking for a paradigm shift then this isnt it)