How a tree becomes a table
JSON can nest without limit while CSV is flat, so something has to give. Nested objects become dotted column names, and arrays are squeezed into a single cell:
{"id":1,"address":{"city":"Boston"},"tags":["a","b"]}
id,address.city,tags
1,Boston,a; b
A simple array is joined with a semicolon and a space. An array that holds objects is kept as a JSON string in its cell, because splitting it would need extra rows, and that would duplicate the parent’s data.
Records with different keys
Real data is rarely uniform. Columns are the union of every key found across all records, in the order they first appear, and a record without one of those keys gets an empty cell. An empty cell therefore has two possible meanings: the key was missing, or its value was null. If that distinction matters, keep the JSON as your source of truth.
Quoting rules
The output follows RFC 4180. A value is wrapped in double quotes if it contains a comma, a double quote or a line break, and any double quote inside it is doubled. So say "hi" is written as """hi""" with an extra pair around the word, and lines end with CRLF as the standard specifies.
Opening the result in Excel
The Download button adds a UTF-8 byte order mark so Excel displays accented letters and non-Latin scripts correctly. Text copied from the box has no such marker. Regions that use a comma as the decimal point often expect semicolons between columns, in which case use Data > From Text/CSV and pick the comma delimiter.
Excel also rewrites some values on opening. Long ID numbers lose digits past the fifteenth, leading zeros vanish from ZIP codes and phone numbers, and strings like 3-4 may turn into dates. Importing those columns as Text avoids it.
Treat exported data from strangers with care. A cell beginning with =, +, - or @ can run as a formula in a spreadsheet, an attack known as CSV injection. This converter leaves values unchanged so numbers such as -5 stay correct.