Split a CSV by rows, parts, size, or column value
Create fixed-row batches, balanced parts, approximate-size files, or one file per column value. Every output keeps the header, and one ZIP collects all parts locally.
Choose your CSV
Configure the result
Up to 100 outputs. ZIP creation and splitting use browser memory; this is not a streaming large-file service.
Preview
From source file to checked output
- 1
Open the CSV and confirm the detected columns.
- 2
Choose rows per file, number of parts, approximate size, or a grouping column.
- 3
Review the first part, download individual CSVs, or get every output in one ZIP.
Create one inventory file per status
Input
sku,status
A-1,active
A-2,draft
A-3,activeExpected result
active.csv → A-1, A-3
draft.csv → A-2Boundary: Column-value splitting preserves each original row and groups blanks separately. It does not merge values, sort rows, or create more than 100 output groups.
Useful before it becomes a workflow
These tools target repetitive catalog and operations tasks. Today each action is free and local. Reusable mappings, batch execution, automation, and APIs will only be built when repeated demand is proven.
Always validate output against the destination system’s current specification. This site is independent and is not certified by a marketplace.
Before you process a file
Is the header included in every file?
Yes. Every generated part includes the same column header row.
What happens to blank values when splitting by a column?
Rows with a blank value are preserved in a separate file named blank rather than being discarded.
Is approximate-size splitting exact?
It measures UTF-8 CSV bytes including a repeated header and keeps rows intact. A single row larger than the target remains in its own oversized part.
Does the ZIP upload my data?
No. The archive is generated in your browser after the CSV parts are created, and individual CSV downloads remain available.
How many rows can I split?
The practical limit depends on browser memory. This MVP accepts files up to 50 MB and warns before processing larger inputs.