If you have two HubSpot contact records for the same person, a fresh import will not clean that up. Import updates work by matching an incoming row to one existing record and overwriting mapped fields on it — they do not delete records. The extra record just sits there. So the choice between merging and correcting your import file is not really a style preference; it depends on what your database looks like before the file lands.

When the import alone is enough

Import deduplication works cleanly when exactly one record already exists for each entity. For contacts, include the Email column and HubSpot will match the incoming row to the existing contact and update only the mapped properties. For companies, include Company domain name. For any object, you can export your records, grab the Record ID column, add it to your import file, and use it as the unique identifier — rows with a Record ID update the matched record; rows without one create new records.

The practical workflow for a mixed file is to split it: one file with Record IDs for existing records, one without for net-new ones, then import both separately.

One thing worth checking before you rely on Record IDs: if the file originated outside HubSpot — from an API integration or a third-party sync app — the domain-name deduplication may not have fired on the way in, so you may already have duplicates you haven’t noticed. It’s worth searching for existing records before you import an external file.

When you have to merge first

If multiple HubSpot records already share the same email address, an email-based import cannot update any of them. HubSpot will throw an error and skip that contact entirely. The row is not imported in that email-match scenario. You have to resolve the duplicate records in HubSpot before the import can do anything useful with that contact.

The same logic applies to companies if you are using company domain name or a custom unique-value property: duplicates block the match; the import does not resolve them.

The decision in one rule

Correct the import file when one target record exists per entity and your file can point to it with a unique identifier. Merge first when duplicate records already exist in HubSpot for the same entity, because the import may skip the contact on error and will not remove the extra record.

If you are unsure which situation you are in, export the relevant object, filter for email addresses or domain names that appear more than once, and merge those before touching the import file.