During my 30+ years of computing, I had managed to lose data in many creative and exciting ways. I had discovered that files could contain nothing when Windows said they contained something; I had found that one disk drive was not able to read data saved on another disk drive; I had wiped the wrong disk. No matter what the unpleasant backup surprise might be, I had probably contributed actively to its existential manifestation in the known universe.
After possibly exhausting my usefulness as a source of amusement, the gods deigned to grant me a speck of enlightenment, in the form of the Beyond Compare and DoubleKiller programs. There were probably a thousand ways to use these programs that I had not yet figured out, but I did manage to develop one way of using them that I found helpful.
My way of using Beyond Compare was to set it up as a comparison of the contents of two disk partitions. The source was on the left; the target (i.e., backup disk) was on the right. Keeping it organized that way helped to reduce my risk of accidentally backing up the target onto the source, which was one more way in which I had provided cosmic entertainment at one time or another.
Beyond Compare seemed to want me to set up different workspaces and, within a given workspace, it seemed to want me to set up different tabs, each of which would have one of those source-target pairings. So, for example, I might have a tab for drive D and a tab for drive E, and might save those two tabs in a workspace called Daily Backup. Each of those two sessions (i.e., tabs) could have its own comparison settings. One could conduct a careful byte-by-byte comparison of files on the source and target, which could take more time and demand more system resources; the other might be happy with a quick check of file dates and sizes. In the worst case, some system-hogging configurations encouraged me to try running Beyond Compare in a virtual machine, where I could control its access to drives and the CPU, so as to prevent it from forcing my other computing activities to a standstill.
Each time I would run Beyond Compare, I would see new files on the left side. Some of those files would be newly created; others would have been moved from folder A to folder B. My approach was to copy all new files from the left (source) side to the right (target) side, but making sure that I had not marked anything for deletion on the right side. When that was done, the target would have everything that was on the source, plus some duplicates (e.g., a copy of the same file in both folder A on the source drive, where it was before, and folder B on the source drive, where it had been moved).
This is where DoubleKiller came in. After copying those files from source to target in Beyond Compare, I would run DoubleKiller on the target drive. It would detect that the backup now contained a copy of file X in folder A and also in folder B. DoubleKiller would let me choose one or the other for deletion. To make that choice, I would rely on what I was seeing in Beyond Compare, and also on what I would remember or could quickly find out using the Everything file finder, to figure out which duplicates I should delete from the target drive. There were also other ways of using these tools and techniques to identify duplicates.
This system had the advantage of offering a hands-on check of what was being saved and not saved in my backups, without the kind of time investment that a fully manual check could take. There would still be an investment of at least five to ten minutes per day, and that could stretch into hours if I was reviewing a cumulative backup of multiple changes stretching back over weeks or months. But the time investment was worth it. In the years since I had adopted this system, data loss had become much rarer. Usually the backup process was mundane, but once in a while something would be changed that was not supposed to change, or something was deleted that was not supposed to be deleted, and I seemed to be catching those.