How to deduplicate a CSV in Excel without losing the evidence
A review-first Excel workflow for finding repeated CSV records, deciding what to keep, and exporting a verified copy without overwriting the source.
Before you work through this by hand: you can check your own CSV free and see which of these problems your file actually has — duplicate rows, near-duplicates, broken quoting, encoding damage. It runs in your browser tab, nothing is uploaded, and there's no signup or email.
Excel can find repeats; it cannot decide what they mean
Excel's Remove Duplicates command is useful for a clearly defined, disposable list, but it removes rows according to the columns you select. It cannot know whether two rows are duplicate contacts, separate orders, or a newer record that conflicts with an older one. Treat its result as a decision you have reviewed, not as an automatic cleanup rule.
1. Preserve the source and define the identity key
Save the original CSV unchanged and make a dated working copy. Decide which fields identify one record: an email, customer ID, order number, or a documented combination. Do not use a shared company name or email domain as the only key when multiple legitimate records can belong to that organization.
2. Normalize only the matching columns
Make a helper column that trims accidental spaces and standardizes case for the matching key. Keep the original value beside it for review. Treat blank keys separately: grouping every blank email or customer ID together creates false duplicates. Preserve identifiers as text so Excel does not remove leading zeros.
3. Highlight groups before removing anything
Use Conditional Formatting or a PivotTable to count repeated keys, then filter each group and compare the surrounding fields. If rows disagree on a phone number, address, status, or timestamp, flag the group for a human decision instead of keeping the first row by default. The conflicting-data guide lists defensible keep rules.
4. Export a new file and verify it
After review, copy approved rows to a new workbook or use Remove Duplicates only on the working copy. Record the original row count, duplicate groups reviewed, and final row count. Export to a new CSV, reopen it, check the header and a few quoted or non-ASCII fields, and confirm the output still imports as the expected number of columns. Never overwrite the source until the destination import has been checked.
When a local review tool helps
For large or messy exports, CSV Repair & Deduper keeps the original values visible while it flags exact and near-duplicate candidates, supports review and undo, and exports a separate cleaned file locally in your browser. Try the free check on your own CSV first; it does not upload the file.
See also: the backup/versioning guide and sample-first testing guide.
See CSV Repair & Deduper (offline browser tool) — $19 one-time