Key takeaways
- Back up everything before you change anything.
- Remove dead records and duplicates first; every later step gets easier.
- Move only the fields someone will use in the new system.
- Opt-outs and consent records must arrive intact.
What should you clean first in a CRM data cleanup
Clean dead records and duplicates first, because every later step is easier with fewer records. Dead records include bounced emails, people who have left their company, companies that have closed and contacts with no activity for years.
Archive rather than delete where you can. Keep an export of what you removed, so history is still available if someone needs it, and check your own retention rules before deleting anything for good.
How do you dedupe a CRM before migration
Dedupe by agreeing a match rule, then merging in batches with a person checking the uncertain pairs. Email address is a strong match for contacts and web domain for companies; a match on name alone needs a human look.
Decide which record wins before you start, such as the one with the most recent activity or the most complete details, and make sure notes and activity from both records are kept.
How do you map fields for the import
Map only the fields someone will use in the new system, and write the map down before you import. A field map is a simple table: old field, new field, type, and what happens to the values.
- List every field and how often it is actually filled in.
- Drop or merge fields nobody fills in or reads.
- Standardise choice values, such as country names and stage labels.
- Turn free text into fixed choices where the values repeat.
- Match field types: dates to dates, numbers to numbers, yes or no to yes or no.
Why do owners and consent records need special care
Owners and consent records need care because losing them causes the most visible problems after go-live. A record with no owner gets no follow-up, and records owned by people who have left need reassigning before the move.
Consent matters even more. Carry opt-outs, unsubscribe flags and any notes on the basis for contacting someone across intact, because a lost opt-out can mean emailing a person who asked you to stop. Under GDPR, someone who objects to direct marketing should not receive it again, and the mistake also costs trust.
How do you prepare CRM data for import into a connected setup
Do the bulk move through the new CRM's own import, and expect connected tools to update it one approved write at a time afterwards. A CRM reached through MCP is not filled by a bulk sync.
That makes the clean-up before the import more important, not less. Assistant-driven updates are good for keeping records in step day to day, but they are not the way to move or repair thousands of records at once.
How to clean a CRM before migrating, step by step
Work in this order; each step makes the next one smaller.
- Export a full backup of every object, including notes, activity and opt-outs.
- Archive dead records and merge duplicates using an agreed match rule.
- Write a field map that lists each field, its new name and type, and the fields you drop.
- Give every remaining record an owner, and reassign records of people who have left.
- Check that every opt-out and consent record has a field to land in.
- Import a small sample, check it record by record, then import the rest.
Where Pebbles Ai fits
- CapabilityProspecting & customer recordsThe Pebbles Ai CRM (in Beta) imports contacts from Google, Microsoft or a CSV, merges 2 to 100 records at a time with restore from the Bin, and holds admin-defined custom fields of Text, Number, Yes or no, Date and Choice.
- FoundationIntegrationsIf you move to Attio or Pipedrive, the assistants can read records there and create or update them through MCP from chat, one approved write at a time under the execution mode. It is not an automatic or two-way sync.
Common questions
Should you clean CRM data before or after migrating?
Before. Cleaning in the old system means you import less, map fewer fields and start the new system with data people trust.
Is it safe to delete old CRM records?
Archive them in an export first and check your retention rules. Keep opt-out records, even for people you no longer contact, so they are never emailed again.
Who should own a CRM migration?
One person, often in revenue operations or the CRM administrator, with a sales lead who signs off the field map and ownership rules.
Can AI clean CRM data for you?
An assistant can help spot likely duplicates or fill a field one record at a time, with approval. The match rules and the decisions about what to drop stay with a person.
About this page. Written by the Pebbles Ai team for people who run sales, marketing and growth. Last updated .
