Why did my Attio CSV import create duplicates?

Because no unique attribute was mapped. Attio uses unique attributes — most commonly email address for people and domain for companies — to check whether a record already exists, and updates it rather than creating a duplicate. Import into People or Companies without mapping one and Attio has nothing to match on, so every row becomes a new record.

How do you prevent duplicates when importing a CSV into Attio?

Map a unique attribute in the column mapping step: email address for people, domain for companies, or record IDs when you are updating records you exported from Attio. When importing into a list and updating multiple entries that share a parent record, map entry IDs instead — unique attributes alone cannot match those reliably.

Can you undo a CSV import in Attio?

There is no single undo for an import. Records created by a bad import have to be found and removed or merged individually, which is why the mapping step deserves more attention than it usually gets. If the duplicate records carry no activity yet, deleting them is reasonable; if anything has happened on them, merge instead so the history is preserved.

Does re-importing a corrected CSV remove the duplicates?

No. Attio's importer creates and updates records; it never deletes them. Rows you removed from the file simply do not appear in the import, leaving the records behind them untouched, and rows whose identifying value you corrected no longer match anything and are created as new records.

Duplicates from a CSV import into Attio

Last checked September 1, 2026 · Deduptio is our product. Every claim about another tool comes from its public site or documentation and is linked in place — tell us if something is out of date.

The short answer

Map a unique attribute — email for people, domain for companies — and Attio updates the record that already holds that value instead of creating a second one. Skip that step and every row becomes a new record. There is no undo for an import, so the mapping screen is the whole game.

How Attio decides to update or create

Attio's importer checks incoming rows against existing records using unique attributes. If a row's value matches an existing record, that record is updated. If it matches nothing, a record is created. Email address is the usual unique attribute for people and domain for companies; record IDs work when you are re-importing data you exported from Attio.

There is no third option. The importer never deletes, which is the fact behind almost every surprise on this page.

The two mapping mistakes

Mistake	What you get	The fix
No unique attribute mapped	Every row is created. Import 3,000 contacts you already have and the object doubles	Map email address for people, domain for companies, before running the import
Importing into a list without entry IDs	The same record added to the list more than once — one record, several entries, with counts to match	Map entry IDs when updating multiple entries that share a parent record — see duplicate list entries

Before you import: five minutes that saves an afternoon

Deduplicate the file itself. Two rows with the same email will be applied in sequence — the second one overwrites the first. This is the one place a spreadsheet is genuinely the right tool.
Normalize the identifying column. Lowercase and trim emails. Strip https://, www. and trailing slashes from domains. A value that looks equal to you may not be equal to a matcher.
Fill the identifier or accept the consequence. Rows with a blank email or domain cannot match anything and will always create. If a third of your file has no identifier, a third of the file is new records.
Import companies before people. Associations resolve against companies that already exist; the other order creates company records as a side effect.
Test with twenty rows. Import a slice, look at what happened in Attio, then run the rest. This catches every mapping error at a cost of twenty records.

If the import has already run

Do not re-import a cleaned file: it cannot delete anything, and correcting an identifying value while cleaning turns an update into yet another new record. Work on the live records instead.

Scope it. The duplicates are the ones created in the import window and they share the shape of the file, which makes them unusually easy to match on the exact attribute you failed to map.
Decide delete or merge. If the newly created records have no activity yet — nothing emailed, nothing noted — deleting them is clean. If anything has landed on them, merge so the history moves onto the survivor rather than being destroyed.
Choose the survivor deliberately.Usually the original record, which carries the history. The imported one often has better field values, so copy those across or let the merge's field rules do it.
Clean the lists afterwards. If the import also added entries, merging the records leaves the survivor in the list twice.
Re-check counts. Compare the object count against what it was before the import. That number, not a spot check, tells you the cleanup is complete.

Doing the cleanup with Deduptio

An import-created duplicate set is the easiest kind to clear because the match is exact: two records, same email or same domain, one created minutes ago. Run an exact rule on the object, dry-run it, check a sample, and merge the batch. Every merge is snapshotted and restorable for 24 hours, and duplicate list entries left behind are cleaned automatically.

Find out what the import created

Connect your Attio workspace, write one match rule, and run a scan. Scans are read-only — you see every group and the reason it matched before anything is merged.

Start a free Attio duplicate scan

Duplicates after a HubSpot to Attio migration — if this import was part of a move, read that one instead.
Why a cleaned CSV cannot fix Attio · The full guide