Nested JSON to CSV: arrays, missing fields, and spreadsheet checks

Decide what one row is supposed to represent

An API can return a customer object inside an order and several tags inside an array. A spreadsheet needs a column for each value. Before exporting, decide whether one row represents an order, a customer, or an item. Flattening the input cannot make that choice for you.

Utilza's conversion uses one row per top-level object. A nested object can become dot-path columns such as customer.city. An array inside that object stays in one cell as JSON text. It does not create a new row for each tag or item.

Keep the JSON source alongside the CSV. An explicit null and a missing field both become empty cells, so the table cannot tell them apart afterward. If that distinction affects a report, add an explicit status field in your export process before converting.

Convert nested JSON to CSV and inspect what changed

Open JSON Viewer to inspect the top-level shape and locate the records you need. If the response is an object containing an orders array, copy that array when you want one row per order. A top-level wrapper object otherwise becomes a single row. If parsing fails, follow the guide to finding the first invalid JSON error before converting.

Then open JSON to CSV, keep Flatten nested objects and Include header row enabled, and choose a delimiter. This compact input shows the two different kinds of nesting:

[
  {"id":"00123","customer":{"city":"Lisbon"},"tags":["new","retail"],"optional":null},
  {"id":"00456","customer":{"city":"Paris"},"tags":[]}
]

The resulting columns are id, customer.city, tags, and optional:

id,customer.city,tags,optional
00123,Lisbon,"[""new"",""retail""]",
00456,Paris,[],

The doubled quotes in the tags cell are CSV escaping. After a CSV parser reads the cell, its value is ["new","retail"]. Both optional cells are empty even though the first input explicitly contained null and the second omitted the key.

A local fixture run on October 27, 2026 used Utilza's current JSON parser and CSV helpers in Node.js v24.15.0. It checked nested objects, string arrays, missing keys, null, 00123, a comma, an embedded quote, a line break, and =1+1. The raw CSV retained the leading-zero text and quoted the comma, quote, and multiline fields. This tests the converter's output; it does not prove how every spreadsheet will import those cells.

Before opening untrusted records in a spreadsheet, enable Excel-safe. In this tool it adds a UTF-8 byte order mark and prefixes cells starting with =, +, -, @, a tab, or a carriage return with an apostrophe. The fixture's =1+1 became '=1+1. That prefix changes the stored cell text. Keep formulas from unknown exports disabled.

Import identifiers such as ZIP codes and product codes as text in your spreadsheet. CSV quoting, including "00123", does not by itself tell the spreadsheet that a column must remain text. Use its import controls to choose the delimiter and column types.

You can pass the CSV through CSV to JSON to inspect the table again. With Convert types enabled, the tested 00123 stayed a string and empty cells became null. The key customer.city stayed a literal key, and the tags cell stayed a string. Reversing the file format does not rebuild the original hierarchy.

The converter's quoting follows the rules described in RFC 4180: a field containing a delimiter, a quote, or a line break is quoted, and an embedded double quote is doubled. Those rules protect the table structure; they do not preserve JSON types or control spreadsheet formulas.