Edit a large CSV.
Keep the original.

1. Check headers, separators and encoding

Choose Open data and select the file in the native file picker. Review the detected source before opening. If all fields appear in one column, check the separator; if text looks wrong, check encoding. An incorrect header choice can turn the first record into field names.

Inspect a few complete values in free Preview before planning an edit. Quoted commas, embedded line breaks, empty values and long identifiers deserve attention. Follow the large CSV opening guide if the detected structure does not match your source.

2. Locate the intended source record

Find loaded rows searches the page already loaded. Use the supported Pro full-file filter when the record could be elsewhere in the source. Choose a meaningful identifier, category or time interval, inspect the result and add sorting only when it helps the investigation.

Keep the original source-record context with the selected cell. The first row in a sorted or filtered result is not automatically the first row of the original file. A stable identifier is useful for checking the saved output independently.

For a research export, the backtest investigation guide explains how to retain the source population and metric meaning. The example below is synthetic mock data, not a real strategy result.

3. Stage a small, reviewed cell change

1. Switch from Inspect to Edit.
2. Select the field in the intended record and read its complete value in the inspector.
3. Enter the replacement and choose Stage edit.
4. Review the staged change. Use undo if it affects the wrong record or field.
5. Repeat only for the records you have checked.

Scroll horizontally to compare every field. Keyboard: focus the table and use the arrow keys.

For this mock task, only the label belonging to mock_002 should change. Inspect mock_003 afterward as an unaffected record. Enter the intended cell value; CSV delimiters and quote escaping are part of writing the output file, rather than instructions to concatenate fields by hand.

4. Distinguish editing scope from export scope

In the CSV cell-edit workflow, Save copy writes a copy of the source with the staged source-record edits. The current filter or sort helps you find records; it does not automatically make the saved artifact a filtered-only or sorted export.

If you need only a subset or a different output order, establish an explicit supported export workflow for that result. Check the output's row count, identifiers and order. Seeing a narrow grid is not evidence that the saved file contains only those visible records.

5. Save to a new destination and reopen it

Choose Save copy and use the native save dialog to select a separate output path. Keep enough space for the new file. Review progress while the save runs; avoid repeatedly starting the same output operation.

After completion, open the saved file as another source and verify:

• The intended identifier and field contain the replacement.
• Representative unaffected records retain their values.
• Headers, column alignment, delimiter and text decoding match the intended output.
• The saved file's scope and ordering match your task.

Check values that other tools might interpret differently. For example, a mock identifier 000127, a long numeric identifier and a decimal written as 1000.00 deserve a complete-value check in both source and output. A shortened grid label is not enough to verify them. When your downstream task requires the original text representation, compare the decoded fields with an independent CSV reader as well.

The original remains available for comparison. A separate data copy is not a byte-for-byte formatting guarantee: CSV quoting and line-ending representation can be rewritten, and output encoding should be checked when moving between tools.

6. Budget a complete output operation

Even one cell change can require reading and writing the complete source when you save an independent CSV copy. Its duration and disk requirements differ from showing a first page or running a selective filter. Allow space for the output, file indexes and any query work.

The 20 GB mock CSV comparison reports its measured access, filter and sort tasks. Those numbers are not a measured save-copy time. Use the performance model, visible progress and cancellation for the actual operation.

This guide covers the CSV workflow. JSON document editing and bounded Parquet editing have different limits; review the format-operation matrix. See current Pro introductory and renewal terms before purchasing for a repeated task.

Continue your workflow

• Large CSV viewer and editor
• Check source setup and opening
• Measured 20 GB mock CSV tasks
• Use authorized local data tools

Try the first page on your own data.

Start on Windows or macOS: source information and up to 100 records, read-only. Review reader capabilities before planning a full-file Pro workflow.

Version 0.1.1 · Windows x64 · macOS Apple Silicon · Install guide →