Tabular data moves between spreadsheets, databases, APIs and documents constantly, and each hop has its own format. CSV is the lowest common denominator, JSON is what programs exchange, .xlsx is what people actually send each other, and Markdown tables are what ends up in READMEs and wikis. The tools here convert between them without uploading the file — useful when the spreadsheet contains customer lists, salaries or anything else that should not end up on a stranger's server.
CSV is less simple than it looks
There is a standard for CSV (RFC 4180), but real files bend it constantly. Fields contain commas and line breaks and must be quoted; quotes inside quoted fields are doubled; Excel in much of Europe writes semicolons instead of commas because the comma is the decimal separator; files exported with a byte-order mark start with three invisible bytes that break naïve parsers. The converters detect the delimiter, strip the byte-order mark and handle quoted line breaks, and each page lists exactly which variations are supported.
Values that change when you are not looking
Spreadsheet software is eager to be helpful: it turns 007 into 7, 1-2 into a date and long card numbers into scientific notation. When converting to JSON you choose whether values are kept as text or typed as numbers and booleans, and even with typing switched on, values with leading zeros stay as text so identifiers survive. Excel dates are stored as serial numbers; the Excel reader converts them back to calendar dates using the cell's number format, including the historical 1900 leap-year quirk.
Pick the right direction
- CSV ⇄ JSON: either direction, with header detection, type inference, nested-object flattening and optional byte-order mark for Excel.
- Excel to CSV: read any sheet of an .xlsx file and download it as CSV or JSON.
- Markdown table: build or paste a table and get aligned Markdown that renders on GitHub and most wikis.
All three work on the whole file in memory. Very large files — hundreds of thousands of rows — are processed without freezing the page, and the preview shows the first rows so you can check the result before downloading.
Why spreadsheets damage data
Excel is where most CSV files end up, and it changes what it opens: it reads UTF-8 in a legacy code page, drops leading zeros from postcodes, keeps only 15 significant digits of long identifiers, and turns codes such as SEPT1 into dates. The guide to CSV and Excel pitfalls explains each of these, shows how Excel stores dates as serial numbers (45000 is 15 March 2023, or 16 March 2027 in the old 1904 system), and ends with a checklist for moving tabular data between systems without losing anything.
Choosing a format
CSV is the lowest common denominator: every program reads it, but it carries no types, no encoding and no structure. JSON keeps nesting and tells numbers from text, which makes it the better choice between programs. Markdown tables are for people, in documentation and chat messages. The tools here convert between all three, and each page says what is lost in each direction — nested JSON, for example, is flattened into dotted column names such as addr.city when it becomes CSV.