I do it simply. I have a directory on my genealogy disk named Filebox. I make a directory using the RM ID number when I have a document that has a citation for that person, except for census documents. I place a copy of each of those documents in its appropriate folder. For example, I name the files, "ID300DC," ID300stone," etc. for a death certificate and a grave stone. For census images, I create directories for federal and state censuses, each containing specifics. For example, I have Federal> Ohio> Hocking> Falls> Logan for Logan (city), Falls (township), Hocking (county), Ohio (state). The census images for Logan, saved in the Federal> Ohio> Hocking> Falls> Logan directory are named Logan(pagenumber), which corresponds with my citation in RM. For me, this makes reviewing previous research for a particular person simple. I have all the documents cited for a RM ID number in one place, while documents that may apply to multiple people (many of my folks are on one page in the census) are readily recalled.
All these complex systems for storing digital documents get in the way of doing real research. KISS.