CSV is the lowest-common-denominator format for spreadsheet data. Every database import tool, script, and legacy system accepts it, while plenty choke on the .xlsx binary format's formatting, formulas, and multiple sheets. This pulls the data out as plain comma-separated values, ready for a database import or a script.
A spreadsheet is more than its grid
An XLSX is a zip archive of XML describing a workbook: multiple sheets, formulas, cell formatting, column widths, merged cells, conditional colours, charts, data validation rules and sometimes macros. CSV is a single grid of text with no concept of any of that.
So the conversion keeps the values and discards the machinery. Formulas are replaced by their most recently calculated results, which is usually what you want but does mean the output is a snapshot rather than a live model. Merged cells collapse in ways that can misalign a header. Anything conveyed purely by formatting, a red fill marking exceptions for instance, is simply not in the output, because CSV has nowhere to put it.
Dates, and why they arrive looking strange
Excel stores a date as a number counting days from an epoch, with the display format held separately as a property of the cell. Which is to say the cell showing 31/01/2026 actually contains something like 46053. Whether a date survives conversion legibly depends entirely on whether that display format is carried through the export.
If a column of dates comes out as five-digit numbers, that is what happened, and the fix is to format the column as text in the spreadsheet before converting. It is also worth deciding what you want from a multi-sheet workbook before you start, since a CSV holds exactly one grid and cannot represent several sheets at once.