Very close to my own experience. It depends on the library, but in my case I realized there is an underlying pattern: for example patches differentiated by length, and simultaneously by attack type. For example: sustain soft and sustain accent; marcato long and marcato short. Or several types of legatos, differentiated by slurring or re-tonguiging, and so on.Then I realized it made sense to arrange my expression maps for that library based on how the articulations are presented in that library's user interface. When working with that library the arrangement of articulations is consistent between exp map and user interface.
So now, for example, I sort the length in Art I and attack type in Art II, all as attributes. This way I assign note-level articulations using the drop down menu - by first selecting length, then the attack type.
This realization that Art I and Art II are actually presented as a drop down and can be chosen sequentially was an eye opener...