Clean a CSV file by checking its headers and records before removing duplicates or merging it with another export. A CSV is a plain-text table, but its meaning depends on the column names, delimiters and values inside it. A file that opens without an error can still contain misplaced fields, missing identifiers or inconsistent records.
The CSV Viewer and Cleaner, CSV File Merger and CSV Comparison Tool support different stages of the job. Keep the original exports and work on copies. That makes it possible to investigate a change rather than wondering what the cleanup removed.
Inspect headers and a sample of rows first
Open CSV Viewer and Cleaner, paste the CSV and choose View. The result reports column names, a row count and parsed records. Look at the first few rows and at records containing commas, quotes or unusual values. Those are good places to detect an incorrectly prepared export.
The first row should contain clear, unique headers such as id,name,status. Make sure each header describes the values below it. Leading zeros in an identifier can matter: 001 is not necessarily the same business value as 1. These tools read CSV cell values as strings rather than deciding which columns should become numbers.
A quoted field can legitimately contain a comma. For example, a company name containing a comma belongs in one properly quoted cell. Do not remove quotation marks simply because they look untidy. They can be part of the structure that keeps one field from being mistaken for two columns.
Choose a duplicate rule deliberately
In the deduplicate operation, an empty column selection compares complete rows. Selecting specific column names compares only those fields. That is a useful distinction: two records can share a customer name while representing different orders. Removing one because the name matches would discard a genuine transaction.
Suppose your file contains id,name,status and the same complete row appears twice. Comparing all columns removes the repeated row. If two rows have the same id but different status values, comparing only id keeps the first occurrence and removes later matches. Use that rule only when keeping the first record is actually appropriate.
Duplicate removal is not an automatic latest-record selection. The tool does not inspect dates and choose the newest row for you. If recency matters, review or order the records using a suitable workflow before deduplicating. Keep a record of the columns used so someone else can reproduce the cleanup.
Keep the columns needed for the next step
The Columns operation creates an output containing the named fields you retain. Enter exact header names separated by commas. Check spelling and capitalization; an unknown header produces an error. Do not drop the unique key column if you will later need to compare records or connect the export to another system.
A focused output can make a review easier. For example, keep id,status,last_contact when those are the fields needed for a follow-up check. Save the full original separately. A smaller table is easier to read, but omitted columns cannot support questions that require their values.
Change delimiters without breaking quoted values
CSV Viewer and Cleaner can export with a comma, semicolon or tab delimiter. Choose what the receiving application expects. A delimiter is the separator between fields, not a search-and-replace instruction for every character in the content. Proper parsing and export preserve separators that belong inside quoted cells.
If you copied a table from a spreadsheet, use Table to CSV Converter for tab-separated input. Include the header row and paste the full rectangular selection. Then check the resulting CSV in the viewer before importing it elsewhere. A visually aligned paste is not enough to confirm that every row has the right fields.
Merge compatible files by appending records
CSV File Merger combines files with identical ordered headers. A file with id,name,status is not compatible with another containing name,id,status, even if the same names occur. Reorder the columns first or export the sources consistently. The tool rejects incompatible ordered headers rather than guessing which fields should match.
Merging appends rows. It does not join related records by an identifier or automatically remove duplicates. For two files containing three and four records, an ordinary append produces seven records before any separate duplicate cleanup. Check totals, then apply a duplicate rule only when the data requires it.
Compare versions using one unique key
CSV Comparison Tool needs a key column present in both inputs. Each key must be nonblank and unique within its file. A stable order number or customer ID often works better than a name, because two people can share a name and one person’s name can change.
The result reports added, removed and changed records. If id 101 appears in both versions but its status changes from open to closed, it belongs under changed. If id 102 appears only in the second version, it is added. If id 103 disappears from the second, it is removed. Check input order before interpreting those labels.
Duplicate or missing keys are rejected because they make one-to-one comparison ambiguous. Fix the source or choose another appropriate key. Do not assign arbitrary unique values simply to silence the error if those values no longer identify the same records across versions.
Review spreadsheet formula protection
On CSV export, formula-looking cells receive an apostrophe to help prevent spreadsheet software from interpreting them as formulas. This is deliberate, so a protected output cell may not be byte-for-byte identical to the original text. Review such fields when the receiving system expects exact strings, and use its supported text-import settings.
Do not assume CSV is harmless merely because it is plain text. Spreadsheet applications can interpret certain leading characters specially. Keep data exports separate from formulas you intentionally want to execute. The cleanup tools organize records; they do not verify the trustworthiness of a supplier or certify an import as safe for every application.
Check the final output before importing it
Compare the input and output row counts, inspect the selected headers, and confirm that identifiers and quoted text remain intact. Try a small import into the receiving system before sending a large file. File tools allow up to ten files, 20 MB per file and 50 MB combined; pasted text has a 200,000-character limit.
For conversion to structured JSON, read the JSON formatting and conversion guide. If the source is a PDF table, follow the table extraction guide before applying these cleanup rules. The CSV format description in RFC 4180 provides background on quoted fields and record structure.