The paper layout approach never really worked for me, at least in the way that I always understood it. Expecting to take a 2D overhead drawing and translate it into 3D is sorta nonsense in my opinion. It's useful for a collection of ideas and going through the motions of figuring certain things out but in the end I find so many of the best decisions are made on the fly and just screwing around with the geometry in editor to get a sense of what works best and what doesn't for the layout/composition/connections between areas.
I have no real good advice for the best process of getting a level going, but I always find the seat-of-your pants approach works best for me. Which probably stems mostly from the fact that I can't draw for shit, so the editor has always been the best tool for me. All I'd say is create one small section of your level that encompasses some of the main ideas you want to include regarding gameplay/theme/architecture/shape/lighting whatever and expand from that one piece. That's essentially how I've done every level I've made (and still do).
Try not to think about it and just get something on screen that looks cool and go from there. I always felt like I was pissing away time over thinking a layout or idea, when I could have just been rolling with it in 3D.