A general principle when working with digital files is to save frequently. When working in ArcGIS, QGIS, or working with geospatial files generally, another good idea is to iterate frequently. Saving frequently will help preserve work in case of crashes or power outages, but if you make a mistake and accidentally erase needed data, saving will do little to prevent that, unless you have been saving multiple copies of the file.
Usually, whenever I am making a change to a file that has not been changed in a while, or whenever I am making a significant change, I will make a copy, append _v2 to the name, and work with that file_v2.shp. This way, if something goes terribly wrong while I am working on complicated_border_analysis_v5.shp, I can load up _v4, instead of having to recreate the whole thing from scratch.
Another measure that makes it easier to recover from errors is using Dropbox. This week, I made a copy of a file that had lots of data, of which I needed three attributes. I then deleted all but those three attributes from the copy. Except I copied the wrong file and then actually deleted the attributes from the original file. It was not a disaster, though, because these files were saved on Dropbox, which has version recovery: it just took a minute or two to remove it from ArcGIS, look through Dropbox’s event history to find the file.dbf, file.shp, and file.shp.xml, revert them back to the previous version, and load that in ArcGIS to verify that all the data had been restored.
It is less about the specifics of whether you use Dropbox, GeoGit, OneDrive, Git, or some other cloud tool, or whether you manually or automatically iterate files, and more that you have thought out a plan to deal with these issues. Because that forethought can save a lot of annoyance later on.
maybe its a map 