Both directions, in this tab. Your file never leaves it.
It will not turn your postcodes into numbers.
The classic way a spreadsheet tool destroys data is by being clever: it sees
007 and stores 7, sees SW1A 1AA and mangles it, sees a
date and reformats it to whatever locale it feels like. Anything with a leading
zero is treated as text here, because a value that starts with a zero is nearly
always an identifier — an account, a phone number, a postcode — and
never a quantity.
CSV is read with its quoting rules honoured, so a comma inside a quoted field
stays inside that field. The delimiter is detected rather than assumed, so the
semicolon-separated exports that most of Europe produces work without a setting.
1. Open a file
Drop a .xlsx or .csv here, or click to chooseNothing is uploaded — it is opened inside this tab
2. What is in it
3. Save it
The honest notes
Nothing is uploaded. Static page, no endpoint, no account. An .xlsx is a ZIP of XML and the browser already has a ZIP engine, so this half needs no download at all.
Values, not formulas. A cell containing a formula exports the value it last calculated, which is what the spreadsheet was showing. Formulas themselves are not carried into CSV, because CSV has no way to hold one.
Formatting is not carried across. Colours, fonts, column widths, merged cells, charts and conditional formatting are presentation, and CSV has none of it. Going the other way, a CSV becomes a plain grid.
Dates come out as the file stored them. Excel keeps a date as a number with a display format attached; without applying that format the underlying number is what you get. Reformatting dates by guessing is how tools turn 03/04 into the wrong month, so this does not guess.
Every sheet is read. Pick which one to export; CSV holds exactly one grid, so one sheet at a time. Exporting to .xlsx keeps them all.
Big files are real work. Everything is held in the tab, so a spreadsheet with hundreds of thousands of rows will be slow on a phone.