Choose the exports
Select Two files and choose the baseline and new .csv or .tsv files. Direct file mode can compare a CSV with a TSV. Start with Delimiter: Automatic and Text encoding: Automatic under CSV and TSV options.
These files contain text values and table structure. They do not carry Excel borders, fills, images, or workbook formula objects. A formula-looking string in a CSV is text.
Tell the app how to read the file
| Control | What to choose |
|---|---|
| Delimiter | Automatic, Comma, Tab, Semicolon, or Pipe. If a whole row seems to be one field, check the separator. Quoted separators, escaped quotes, and quoted multiline fields are supported. |
| Text encoding | Automatic, UTF-8, UTF-16 little-endian, or Windows-1252. If accented names look wrong, choose the encoding used by the export. |
| First row contains column names | Leave on for a header such as Record ID, Name, Amount. Turn off if the first row is already a record. With headers off, refer to columns by their one-based numbers. |
| Match rows by | Row position compares rows in their existing order. Key column matches records using a unique ID; see the separate key guide. |
Ignore columns that always change
In Ignore columns, enter comma-separated header names or one-based numbers: Last Updated, Notes, 7. Names are matched without case differences. Use numbers when headers are off or when a header name appears more than once.
Named ignored columns must be in the same position in both files. This option skips the relevant value and header checks; it does not promise to hide every structural change associated with a column.
In the inventory exercise, compare the two copies of 05 Inventory.csv by Row position. With no ignored columns, there are six differences. Ignore Last Updated and rerun: two findings remain—a chair quantity changes from 18 to 16, and a new task lamp row appears with 22 units.
Numeric-looking text
By default, 001 and 1 are different strings. That is often correct for identifiers. A positive Numeric tolerance enables numeric comparison when both values can be read as supported numbers. Be careful when your file mixes amounts with codes that happen to contain digits.
If reordered rows produce a wall of differences, use Match rows by Key column. It is designed for that situation; ignoring lots of rows is usually the wrong fix.
Still have a question?
Tell us what you are trying to check and where you got stuck. Contact us.