Let me add more information -
For example, I have a document graphic that seems to be folded in its edge and is filled with text, and I want to use the same document but without the fold and without the text on it. I have to duplicate the composition and change it.
This adds up very quickly, despite the fact both animations share some nested comps and look quite similar..
Another more complicated example -
I create a scene of 2 characters (pre-rigged, pre-comped, shared through all of my videos) walking down the street and then shaking their hands.
Later on, I need a similar scene with 2 characters walking down the street, waving, and then shaking their hand.
I will then duplicate the parent comp, duplicate the characters comp, slightly change the animation of the characters to wave before they shake, and use it.
Now I have multiple comps which are quite similar and share some comps (like the street comp), but have different nested comp
I hope that explains the situation better