CSV is where data arrives. Parquet is where it should live once you start querying it. This converter reads your CSV, works out what each column actually contains, and writes a typed, compressed Parquet file — without the data leaving your machine.
Reading CSV
Auto-detect reads the first rows and picks the separator that splits them consistently.
Turn off when the file starts straight into data. Columns are then named column_1, column_2, and so on.
Turns "42" into a number and "true" into a boolean. Values with leading zeros stay text so IDs and zip codes survive.
Which worksheet to convert. Filled in once the workbook is read.
Dotted path to the array of records, such as data.items. Leave empty to use the whole document.
Turns {"user":{"id":7}} into a user.id column. Arrays are kept as JSON text.
Pick the encoding the file was written in. Wrong encoding shows up as garbled accents.
Comma-separated words that should become null instead of text.
Writing Parquet
Snappy is the default across Spark, DuckDB and pandas. Gzip is smaller but slower to read.
Turns a user.id column back into {"user":{"id":7}}.
Excel needs a byte order mark to open UTF-8 files with accents correctly.
Rows per row group. Larger groups compress better; smaller groups let readers skip more.
conversion-preroll
This short ad is what keeps the converter free.
Parquet is unreadable in a text editor. That is by design.Check your new Parquet file with ParquetReader — opens Parquet files directly.Plain text rows that open anywhere, with no types and no schema. Typed, compressed, columnar storage built for analytical scans.
| CSV | Parquet | |
|---|---|---|
| Type information | None — every value is text | Full — ints, floats, booleans, timestamps, decimals |
| Structure | Flat rows and columns | Columnar, with nested type support |
| Compression | None | Built in (Snappy, Gzip, Zstd) |
| Typical file size | Baseline (100%) | Typically 5–15% of the CSV |
| Schema | Header row at best | Stored in the file footer |
| Human-readable | Yes, in any text editor | No — binary |
| Tool support | Universal | Spark, DuckDB, pandas, BigQuery, Athena |
Type inference looks at every value in a column, and a single "N/A" in a numeric column forces the entire column to text. Add such words to the "treat as empty" list so they become nulls and the column stays numeric.
A column of "2024-01-31" is written as text, not as a Parquet timestamp. That is deliberate: guessing between day-first and month-first formats corrupts data silently. Cast the column in your query engine after loading, where the format is explicit.
Snappy decompresses fast and is the ecosystem default. Gzip produces a noticeably smaller file but costs CPU on every read. Pick Gzip for cold archive storage and Snappy for anything you query regularly.
Both read Parquet far faster than CSV because they can read only the columns a query touches, and skip row groups using the statistics stored in the file.
A typed, compressed Parquet file is commonly 5–15% of the size of the CSV it came from. On S3 or GCS that difference shows up directly on the invoice, and again in every scan-based query charge.
A CSV forces the next person to re-guess every type. Parquet stores the schema in the file, so the column that was an integer for you is an integer for them.
sticky-bottomYour file is downloading
post-conversion