How to Compare Two CSV Files

Quick Answer

To compare CSV files, paste the original into the first pane and the changed version into the second. The diff matches rows by the first column (when it is a unique key on both sides) and reports what was added, removed, and changed, down to the individual cell: email "old" → "new". Row order does not matter, neither file leaves your browser, and it is the fastest way to find changed rows between two exports.

How It Works

Both files are parsed as real CSV first, following RFC 4180, so quoted fields, embedded commas, and multi-line cells are handled correctly. The comparison then works on rows, not lines, which is what makes this a CSV comparison rather than a text one. When the first column holds a unique value for every row on both sides, it becomes the row key: rows are paired by key, and the paired rows are compared column by column, with columns matched by header name. Rows whose key appears on only one side are reported as added or removed; paired rows with differing cells are reported as changed.

If the first column repeats or is empty somewhere, there is no reliable way to pair old and new versions of a row, so the tool falls back to whole-row comparison and says so in the report. In that mode an edited row shows up as one removed row plus one added row.

Reading the Result

The report opens with the counts: added, removed, changed, unchanged. Then come the details: lines starting with + are rows only in the changed file, lines with - only in the original, and lines with ~ are keyed rows whose cells differ, listing each column as old value → new value. Column-level changes are listed separately by header name, so a renamed column reads as one removed and one added. Remember that identical row counts do not mean identical data: a file where a single cell changed has exactly the same shape, and only the changed-row report reveals it.

CSV Diff vs Text Diff and Spreadsheets

A text diff on CSVs flags reordered rows as changes, because it compares files position by position. Sort one export differently and every line moves, so the diff reports a wall of changes when nothing in the data changed at all. Spreadsheet workarounds, like conditional formatting across two sheets, break down on added and removed rows because everything below the insertion shifts. A key-based diff sidesteps both problems, which is why database tools compare tables this way. For prose or code, where line order is the content, use the text compare tool instead.

Examples

A customer list, before and after. For example: yesterday's export has customers 1, 2, and 3. Today's is missing 2, has a new 4, and customer 3 has a new email address. The report reads: 1 added, 1 removed, 1 changed, with the email shown as old → new. Load the built-in example to see exactly this case. The same pattern works for auditing a price list or inventory file against last week's version.

A re-sorted export. The same rows exported in a different order compare as identical: no differences found. A text diff on the same two files would flag nearly every line.

A schema change. If the new export gained a column, the report names it under "Columns added" and still compares the columns both files share, so a migration that added a field does not drown out the row-level changes.

Common Mistakes