How HubSpot's match logic works

The one rule underneath everything

Contacts are keyed on email address. That single fact explains most of what follows. Two records with the same email are treated as the same person; two records with different emails are treated as different people, no matter how identical everything else looks.

Companies are keyed on domain, which is why two branches of the same business often end up as one record, and why a customer using a Gmail address never associates to anything.

What it will do for you

• Deduplicate on exact email match at import, updating the existing record rather than creating a second one.
• Associate a contact to a company when the email domain matches an existing company domain.
• Surface likely duplicates for review, based on email and name similarity.

What it will not

• Merge two records for the same person with a work address and a personal one.
• Treat sarah@ and admin@ at the same domain as the same person, which is correct, even though the phone number is the same.
• Catch a record created from a phone call with no email at all — the most common source of duplicates in trades businesses.
• Undo a merge. Merges are one way, which is why the rollback file matters more than the merge itself.

Where new duplicates come from

Cleaning the database once does nothing if the thing creating duplicates is still running. In practice it is almost always one of four:

• A form that creates a record before the email field is required.
• A phone or SMS integration creating contacts keyed on number, not email.
• A second list imported from a different tool with its own idea of the customer.
• Someone typing a customer into the CRM while they are on the phone, because searching felt slower.

The order to do it in

Fix the inflow first, then clean, then keep a rollback file. Cleaning before you fix the source means doing it again in three months, and by then everyone has decided the CRM cannot be trusted.