Bucket
|- filename1
|- filename2
|- filename3
|- filename4
|- filename5
|- ...
I'm not joking. This is because I worked in an organisation where directory structures and silo'd access across each department were removed, and everything got thrown into a pool of documents with files that needed to be correctly names, and attached meta data filled out.
Most people hated it.
But I loved it, because it brought a discipline of accurate filenames and attaching correct meta-data. It brought strong disciplines on security/permissions. It brought an audit trail of who had opened or edited what and when.
Whenever I see a department of a company with endless directory upon directory of files with data replicated across them, I know the department has big people trouble - stuff is being done manually and operational controls are weak. Like seeing Excel sheets with merged cells = non-normalised tables that cannot be pivot-tabled = staff don't know basic Excel functionality = stuff is being done manually, time wasted, effort duplicated.
Everything was accessed by search. This also brought strong levels of discovery: Formerly, a document in another department could have been useful to me, but I'd not know of its existence unless I'd caught wind of someone working on this problem before. Now, with correct permissioning, I would see it through search. This saved incredible amounts of time. I also know that what I wrote 10 years ago isn't stuck in a directory that no one knows exists, it is at the fingertips of anyone who needs it.
bin/ - scripts I use
build/ - software I build
doc/
biz/
taxes/
YYYY/
...
budgets/
receipts/
house/
travel/
personal/
notes/
...
projects/
projectA/
docs/
invoices/
contracts/
proposals/
presentations/
...
bin/
src/
repoA/
...
archive/ - old data
people/
johndoe/ - sent to me by someone else
YYYYMMDD/
...
...
...
projectN/
crap/ - Junk organized by date
YYYYMMDD/
...