JSON is great for moving data around, but it's verbose, untyped, and slow to scan in bulk. Converting it to Apache Parquet gives you a compact, strongly typed, columnar file that analytics tools love. This page does the conversion in your browser, with nothing uploaded.
Best results with JSON Lines
The viewer reads both regular JSON and JSON Lines (.jsonl / .ndjson — one object per line). For tabular data, JSON Lines is the most reliable input: each line becomes a row, and the engine infers the columns and types from the objects. A top-level JSON array of flat objects works too.
Deeply nested JSON is read as nested/struct columns; if you want a flat table, use SQL in the viewer to pull nested fields out into their own columns before exporting.
Why convert JSON to Parquet?
- Smaller and faster. Parquet is columnar and compressed — far lighter than the equivalent JSON.
- Typed columns. Numbers, booleans, and timestamps are stored as real types instead of re-parsed strings.
- Analytics-ready. Spark, pandas, and Polars all read Parquet natively and efficiently.
Private by design. Your file is read and converted locally — in your browser on the web, or on-device in the macOS app. The contents are never uploaded to us or to any third party. Read our privacy policy.
How large a file can I convert? The free web viewer exports up to 1,000 rows — plenty to try it out or convert small files. Pro raises that to 100,000 rows in the browser, and the macOS app exports files of any size, from a few rows to hundreds of millions, with the same local-only privacy.
