Start with the receiving system's import instructions. A clean-looking CSV can still use the wrong headers, include a repeated record or hold an identifier that an earlier spreadsheet already changed. Work on a copy and decide which changes the importer actually needs.
Inspect the table before changing it
Open CSV Viewer and check the delimiter, header row and parsing warnings. Look for rows wider or shorter than the header, repeated column names and important values such as leading-zero IDs. Filtering helps inspect a subset, but a hidden column remains in the viewer's download. Hiding personal data in the preview does not remove it from a shared file.
CSV Cleaner accepts pasted CSV or TSV and selected CSV, TSV or XLSX files. A file replaces the pasted text. For a workbook, choose a worksheet by name or position; an empty selection uses the first visible sheet.
Apply only the cleanup rules you need
Header changes, trimming, case changes and empty-cell markers happen before blank-row and duplicate-row removal. The optional empty-column rule follows cleanup. CSV formula protection runs when exporting.
This order has practical effects. Trimming can make two previously different rows identical. Filling empty cells with N/A can make a formerly blank row survive blank-row removal. A column is removable as empty only if no remaining row has a value in it.
Match header names to the importer, rather than changing them to a preferred style automatically. Repeated names receive suffixes, so inspect those names before mapping fields. Leave case changes off for case-sensitive identifiers.
Test a small supplier export
Suppose the source has a padded sku header, two copies of the same item row and a product code 0012. Trim the header and cell edges, enable duplicate-row removal and inspect the change log. The code remains a string, while the fully repeated row can be removed.
Two records with the same SKU but different prices are different rows. Whole-row deduplication cannot decide which price is current. CSV Diff and Change Report can help review changed records by a chosen key, but resolving conflicting business records needs an explicit rule.
Choose the output for its destination
CSV output can prefix formula-like strings, including supported cases after leading whitespace. That changes the stored text. A spreadsheet can interpret or rewrite the prefix differently, so check the values after import. XLSX output stores inert text cells without that CSV prefix; it does not recalculate workbook formulas.
The optional accent repair handles supported UTF-8 text misread as Latin-1, such as é instead of é. It is not a detector for every encoding mistake. For a file refused as UTF-8, re-export from the source with the right encoding rather than repeatedly applying text repair.
Check the complete result
Compare row and column counts with the removals you requested. A terminal newline is a record separator, not an extra data row. Inspect a quoted comma, a line break inside a cell and an identifier before downloading. The preview shows 40 rows; the file contains the full result.
Limits are 20 MiB, 100,000 rows, 500 columns and 2,000,000 cells, with additional workbook-reading limits. Keep the original, cleaned output and a note of the options. For exact identifiers, configure the destination to import text rather than assuming that every spreadsheet will preserve CSV strings automatically.