How to clean CRM data before migration.
A migration should not copy every old problem into a new system. Use the move to decide what matters, create consistent rules, and test trust before launch.
Data migration is not a single import. It is a sequence of decisions about scope, structure, quality, relationships, and accountability.
Old systems often contain duplicate contacts, inconsistent company names, outdated owners, unused fields, and records with no clear purpose. Moving everything preserves uncertainty and makes the new CRM difficult to trust from the start.
The goal is not perfect data. The goal is data that is useful enough for the workflows, reporting, and customer context the new system needs to support.
A four-stage preparation process.
Decide what should move
Separate active, useful records from information that can be archived. More history is not always more value.
Standardise the structure
Agree naming, formats, required fields, and valid values before mapping anything into the new CRM.
Resolve duplicates and gaps
Choose matching rules, decide which source wins, and route uncertain records for human review.
Test and reconcile
Import a representative sample, inspect relationships and ownership, then compare totals before the final move.
Do not clean data without rules.
Cleaning becomes inconsistent when each person makes a different judgement. Write down the accepted formats, duplicate rules, required fields, archive criteria, and source-of-truth decisions first.
Keep an exception list for records that need human judgement. Forcing uncertain data through an automatic rule can create confident-looking errors that are harder to find later.
Questions to answer before importing.
- Does this record still have operational, customer, reporting, or retention value?
- Who owns the record and is that owner represented correctly in the new system?
- Which fields are trusted enough to migrate, and which should be left behind?
- How will contacts, companies, deals, tickets, and activities remain connected?
- What rule determines whether two records represent the same person or company?
- Who signs off the sample import and the final reconciliation?
What happens after launch?
Good data still needs clear habits and ownership to stay useful.