JSON ⇄ CSV
Convert in both directions, flatten nested objects, and download the file — with your data never leaving the browser.
- Free
- Never leaves the browser
- No account
5 columns.
Everything runs in your browser. Load this page, disconnect from the network, and it still works — which is the point when the data belongs to a client.
What the chat window cannot finish.
A converter that handles the cases the quick ones do not: nested objects and arrays flattened to dotted columns, inconsistent keys unioned rather than dropped, commas and quotes escaped properly, and a real file at the end.
For twenty rows, an assistant is genuinely faster and you should use it. For four thousand rows it will not fit, and if it does it may silently drop or invent records — which is the worst possible failure mode for data you are about to act on.
There is also the file. Chat gives you text in a scroll box. This gives you a .csv you can open in Excel, and it does the conversion locally, so client data under NDA never crosses a network.
Bring your own AI
Turn messy text into JSON first
Have a PDF table, an email thread or an unstructured dump? Give your assistant this prompt, then paste the JSON it returns into the converter.
Convert the data below into a JSON array of objects.
Rules:
- Return a flat JSON array. Every object must use the same keys.
- Use lowercase snake_case keys derived from the source labels.
- Keep numbers as numbers and booleans as booleans — do not quote them.
- Use null for genuinely missing values. Never invent a value.
- Preserve every record. If you cannot parse a row, include it with a "_parse_error" key explaining why.
- Return only the JSON, in a single fenced code block, with no commentary.
The data:
[paste your messy data here]We do not call a model on your behalf. There is no API key here and no per-use cost — the assistant you already trust does the language half, and this page does the arithmetic.
Conversion rules
- Nested objects
- Flattened to dotted keys — `user.address.city`
- Arrays of scalars
- Indexed columns — `tags.0`, `tags.1`
- Inconsistent keys
- Union of all keys; missing values left empty
- Escaping
- RFC 4180 — quotes doubled, fields with , " or newline quoted
- Type coercion
- CSV → JSON infers number, boolean and null; everything else stays a string
- Encoding
- UTF-8, with an optional BOM for Excel
These constants are published under CC BY 4.0. Quote them, cite them, put them in your own documentation — attribution to Original Design Studio is the only condition.
The principles behind what this tool tells you.
How do I convert nested JSON to CSV?
Paste it and the converter flattens nested objects to dotted column names automatically — `user.address.city` becomes a column. Arrays of scalars become indexed columns.
Is there a file size limit?
Only your browser's memory. Files in the low tens of megabytes convert fine; conversion runs on your machine, so speed depends on your machine.
Does my data get uploaded?
No. Everything runs in the browser. You can load this page, disconnect from the network, and it will still work.
Why does Excel mangle my accented characters?
Excel assumes the system codepage rather than UTF-8 unless the file carries a byte-order mark. Tick 'BOM for Excel' before downloading.
Next tool
05 — Colour Vision Simulator
Want this run across your whole product?
An audit scores what you have shipped against all ten disciplines — tested with assistive technology and on real hardware — and returns a prioritised plan.