Convert XLSX to CSV

The reason for doing this is nearly always a system on the other end: a database import, a script, an upload form, a tool that predates Excel formats. CSV is the plainest possible table — values, commas, line breaks — and its plainness is exactly why everything accepts it.

Everything a spreadsheet does beyond holding values. Formatting, colours, column widths, frozen panes and cell comments have nowhere to go. Formulas are resolved to their calculated results, which is what an import wants. Merged cells contribute their value once and leave the remaining positions empty, which is worth checking if your sheet uses them for layout.

This is the limit to know before you start. A CSV holds one table and a workbook can hold twenty, so only the first sheet is converted. There is no honest alternative — concatenating several sheets into one file would produce something that looks like data and is not.

If you need several, save them separately in the spreadsheet first and convert each one. It is two extra minutes and it is the difference between an import that works and one that fails halfway through with a column count error.

The CSV is written as UTF-8 without a byte-order mark, which is what every modern importer expects. Excel on Windows is the awkward exception: double-clicking a UTF-8 CSV can turn accented characters into mojibake, because Excel guesses the encoding from the system locale rather than reading it.

The fix is to import rather than open — Data → From Text/CSV, then pick UTF-8. Nothing is wrong with the file; Excel is simply reading it with the wrong assumption.

This runs entirely in your browser. Spreadsheets are frequently the most sensitive files an organisation has — payroll, customer lists, pricing — and there is no reason for one to travel anywhere to have its commas rearranged. It also means no Office installation is required, which is often why the search happened at all.

A spreadsheet does not store a date. It stores a number — days counted from 30 December 1899 — with a display format laid over it, and only the number crosses into the CSV. A cell showing 04/03/2026 is written as 46085. The format that made it look like a date lived in the workbook, and the workbook is what you are leaving behind.

So the surprise is not which way round the day and month are; it is an integer arriving where the import expected a date. There are two ways out and it is worth picking one before converting. Change the column to text in the spreadsheet, so the form you want — usually ISO — is the stored value rather than a costume over a number. Or add the serial to 1899-12-30 on the receiving side, which is what a database import does anyway. Times ride in the fractional part of the same number, so 46085.5 is midday on that day.

A cell displaying 1,234.50 with a currency symbol holds the number 1234.5, and the number is what the CSV receives. Symbols, thousands separators and fixed decimal places are display instructions, not data.

This is almost always what an import wants — a parser reading 1,234.50 as a number will either fail or read 1. If a human is going to read the CSV instead, expect it to look plainer than the sheet did, and know that this is correct rather than lost.

A value containing a comma is written inside quotation marks, so an address held in a single cell survives intact rather than splitting into three columns. A value containing a quotation mark has it doubled, which is what the CSV convention specifies and what every real parser expects.

The awkward case is a cell containing a line break — typically the notes column of an exported form. It is quoted and stays one value, but simple tools that split on newlines instead of parsing properly will still get it wrong. If the import is going to be fragile, strip the line breaks in the spreadsheet first.

What converts is the sheet's used range, which is not always what you can see. A row emptied by clearing its contents rather than deleting it still counts, so a file can end with a run of blank lines.

Hidden rows and columns convert as well, because hiding is a display property and not a form of removal. If a sheet has columns hidden because they were not meant to be shared, delete them before converting rather than trusting the hiding.