I dislike giving wolves priority on principle, but that’s a thing you can only do if players aren’t consciously aware that’s how it works because there are ways to use it for alignment peeks. Wolf RBer A targets RBer B, B targets A, and C tracks A to B, and D tracks B nowhere. That example is contrived to demonstrate the point, but you open yourself up to really strange mechanical peeks if the setups become large enough or complex enough. Randing it is bleh to me because I dislike mechanical nondeterminism. If I had to tell a faction they lost a game because of a cross-targeting coin flip that blows. Like a tied vote rand at least one side can say they should have voted differently or something.
Another solution I don’t use but works is making these actions resolve immediately, and that means they use timestamps to handle cross targeting. That also has a lot of the problems that randing it does where it’s impossible to know what time actions were submitted exactly by anyone but the player themselves (generally).