Type inference: helpful, with one honest caveat

CSV stores everything as text, so the converter infers types: "42" becomes the number 42, "true" and "false" become booleans, "null" becomes null, and anything else stays a string. This is almost always what you want — but note that identifier-like values that happen to be numeric, such as ZIP codes with leading zeros ("02139") or long order IDs, will be parsed as numbers and can lose their leading zero. If a column must stay textual, quote the values in the source or post-edit the JSON output.

Quoting, commas, and multi-line cells

The parser implements the RFC 4180 rules real spreadsheet exports follow: fields wrapped in double quotes may contain commas and even line breaks, and a doubled quote inside a quoted field is a literal quote character. Empty rows are skipped, and a blank header cell is auto-named column_1, column_2, and so on rather than producing an empty key. If your headers use dot notation — user.name, user.address.city — convert first and then run Unflatten in the studio to rebuild the full nested structure.