Agree with the general top-down approach.
Sketch the paper out and leave yourself lots of scaffolding, then make multiple passes adding in detail and refining as you go. Don't be afraid to stop and re-sketch parts if you discover that you've hit a dead-end and can't write yourself out of it.
If you really want to go hard on this approach, use some kind of content versioning system (svn, git, hg) and make a branch for each major expansion, with the ability to discard the branch and return to master at any time. Similarly, you can branch from branches to test something new while you test something new. (Obviously this is way more useful if it's a multi-person project.)
Another topic would be procrastination or (lack of) drive.
Lots to talk about there, but the best strategy I've found is to always get something done every day and never allow yourself to completely ignore the project. Even if what you're doing is an experimental change in wording or styling and you'll throw it away afterward, attempt it.
Best of luck!