File formats matter
What do .csv, .xls, .xlsx, and .ods all have in common?
They're all different ways to represent spreadsheet information. However, they're not all created equal.
I recently had a client that was working on a large analysis piece. He downloaded a data set in a CSV format, and proceeded to add multiple tabs of equations into it over the course of many, many hours. The entire time, he thought that autosave was storing these equations.
He then opens his (still csv) spreadsheet. All of the tabs are missing. He goes to the recovery files. All of the tabs are missing. He goes to his backup. All of the tabs are missing. It is like they never existed in the first place.
The problem is that csv is a very simple format. It doesn't support multiple tabs. It doesn't support much of anything. However, it is really easy to write, really efficient in storage space, and able to be imported into virtually any spreadsheet or database program, so it is a popular format for data sets.
Excel translated this file internally into it's own formats (.xsl or .xslx) which do support these formats. However, the autosave down-converted it back to csv, stripping all of the analysis out of the file.
All of this would have been avoided if he would have initially saved a verson as an .xslx format file, then made the changes.
C'est la Vie.