Deduplicate Excel & CSV files
in seconds

Drop your .csv file into Clean — it runs entirely in your browser, no spreadsheet app, no Power Query, no add-in, no install. Pick the column(s) you want to match on, set a similarity threshold, and download the deduplicated file. Works on Mac, Windows, Linux, and Chromebook, and catches the near-duplicates Excel's built-in Remove Duplicates silently misses — "Jen Walsh" vs "Jennifer Walsh", "Acme Corp" vs "Acme Corporation", and so on. Free for files up to 500 rows, no account required.

Excel's built-in Remove Duplicates only catches character-for-character matches, so "Jen Walsh" and "Jennifer Walsh" survive as two separate rows, and "Acme Corp" and "Acme Corporation" are treated as completely different companies. To catch spelling variants you need fuzzy matching, which scores how similar two strings are between 0 and 1 instead of asking whether they're identical.

We broke this down in detail — including why Microsoft never shipped a real fix and what to do instead — in Fuzzy Matching in Excel (2026): Why It's Still Broken — and the Fastest Fix.

Technically yes — Microsoft's Fuzzy Lookup add-in is still downloadable — but it's a 2017-era Windows-only desktop add-in that hasn't been meaningfully updated, doesn't run on Mac or Excel for the web, and slows to a crawl past a few thousand rows. Clean is the modern browser-based replacement: same fuzzy matching idea, but it works on any OS, handles up to 100,000 rows per file, and supports both .xlsx and .csv. Full breakdown of why the add-in keeps falling short in Fuzzy Matching in Excel (2026).

Power Query's fuzzy merge is the closest thing to a built-in fuzzy matcher in Excel, but it requires Excel for Windows desktop, has limited threshold control, slows dramatically past a few thousand rows, and is really a merge tool — not a dedupe tool. Clean is browser-based, works on any OS, supports up to 100,000 rows per file, lets you tune the similarity threshold, and ships three output formats out of the box (clean file, clusters for review, all rows scored) — none of which Power Query offers natively.

Excel only catches exact character matches — "Microsoft Corp" and "Microsoft Corporation" are treated as completely different records. Clean uses fuzzy matching to score similarity between strings, so Clean catches the variants that exact matching misses. Clean also lets you match across multiple columns simultaneously, so "Jen Walsh at Acme Corp" and "Jennifer Walsh at Acme Corporation" are correctly identified as the same person.

Instead of asking whether two records are character-for-character identical, Clean scores how similar they are — a number between 0 and 1 for every pair. Anything above your threshold gets flagged as a likely duplicate. Common real-world messiness (casing, punctuation, abbreviations, business suffixes like Inc., LLC, and Corp.) is normalised automatically before scoring.

When you select a second column — matching on contact name and company together — Clean combines both signals into one decision. That's how "Jen Walsh at Acme Corp" and "Jennifer Walsh at Acme Corporation" get grouped as the same person, even though neither field is an exact match on its own.

Yes. Select company name and contact name together — or first name + last name + email domain — and Clean combines the similarity across every selected column into a single match decision. Two weak signals (a partial name match and a partial company match) can become one strong duplicate flag, which is how "Jen Walsh at Acme Corp" is correctly grouped with "Jennifer Walsh at Acme Corporation".

For deduplication, 0.75–0.82 works well for most contact and company lists. Go higher (0.88+) if you want to be conservative. Go lower (0.75) if your data is clean and you want to catch more variants.

Unique sheet: one row per entity, duplicates removed. Clusters sheet: only the duplicate clusters with similarity scores, for manual review before merging. All rows scored sheet: your original file with a cluster ID, duplicate flag, and similarity score added to every row.

CSV, XLSX, and XLS. Maximum 10 MB per file. If your file is larger, contact us — we can run it via the API.

Pricing is based on the number of rows in your file, excluding the header row. Clean is free for files up to 500 rows with no account required. For larger files: $1.99 up to 3,000 rows, $4.99 up to 10,000 rows, $9.99 up to 25,000 rows, $19.99 up to 50,000 rows, and $29.99 up to 100,000 rows. You can preview your results before paying — payment is only required to download.

Clean's free tier covers files up to 500 rows. For a 50,000-row file you can preview the deduplicated results for free — see exactly which clusters Clean would flag — and only pay $19.99 to download the cleaned file. If you need to process this volume regularly or larger, the underlying Similarity API handles millions of rows for $1.99 per 10k rows on a pay-as-you-go plan.

Deduplication finds duplicate records within a single file — two rows in the same spreadsheet that represent the same contact or company. Reconciliation compares two separate files — checking which rows in your new list already exist in your reference list, and which are genuinely new. Use Clean when you have one messy file to clean up before importing to your CRM. Use the reconcile tool when you have a new list — a trade show export, an Apollo download, a vendor list — and want to check it against an existing database before importing.

Your file is processed in memory and deleted immediately after your session. It is never written to permanent storage, never shared, and never used for any purpose other than generating your results. You can verify this in our privacy policy.