I use the "Where"(Design Layers)/"What"(Classes) convention described in the video.
Where:
Either specific performance spaces or, if I am using Design Layer ViewPorts, hanging locations or structural elements (like a truss pod). This means that, if I'm not using DLVPs and am working in a single space, I'm working in one Design Layer.
What:
Lighting (conventional, intelligent, notes, dimensions, etc.), audio (PA, monitors, notes, etc.), video (projection, LED, confidence monitors, etc.), staging, scenic, truss, rigging, venue (walls, steel, safety perimeters, etc.), etc.
I end up with a TON of Classes but, with Saved Views, no big deal at all.
I use Classes and Layers similarly to this.
I use layers to broadly categorize my drawings by discipline (one for Theatre Architecture, one for the set, and one for lighting)
I use classes to more specifically organize the file. (lights, positions, labels, dimensions...etc.)
As said above, a lot of classes, but Saved views make it a snap.