CRM Migration Checklist: Move Your Sales Data Without Breaking the Pipeline

Changing CRM is one of those projects that looks like a data export and turns into a quarter of quiet damage. Deals lose their owners, activity history vanishes, duplicate contacts multiply, and three months later nobody trusts the pipeline report. The migration itself is rarely the hard part — the preparation is. This CRM migration checklist walks through the six phases that separate a clean cutover from a slow-motion mess.

Why CRM migrations go wrong

Almost every failed migration fails for one of four reasons:

• Dirty source data. Teams migrate the mess instead of fixing it, then blame the new system.
• Unmapped fields. Custom fields in the old CRM have no home in the new one, so the data silently disappears.
• Lost relationships. Contacts import fine but detach from their companies and deals.
• No validation step. Nobody checks record counts after import, so gaps surface weeks later.

Each of these is preventable with an hour of planning per phase.

Phase 1: Audit before you export

Do not open the export screen yet. First, take inventory.

1. Count everything. Record the exact number of contacts, companies, deals, tasks, notes and emails in the source system. This is your reconciliation baseline.
2. List every custom field, its data type, and how many records actually populate it. Fields under 5% fill rate are usually candidates for retirement.
3. Document your pipeline stages and any automation attached to them.
4. Identify integrations. Every tool writing into the old CRM needs a plan for the new one.
5. Decide what not to migrate. Leads untouched for three years rarely deserve a seat in the new system. Archive them to a CSV instead.

Phase 2: Map your fields

Build a mapping spreadsheet before anyone touches an import button. Three columns: source field, destination field, transformation rule.

• Standard fields — name, email, phone, company — usually map one to one.
• Picklists need reconciliation. If the old CRM has eleven lead statuses and the new one has five, decide the collapse rule now and write it down.
• Date formats are the classic silent failure. Confirm whether the destination expects ISO, US or UK ordering.
• Owner assignment must map to real user IDs in the new system. Create the users first, or every record lands unassigned.
• Orphan fields — anything with no destination — either gets a new custom field or an explicit decision to drop it. No silent losses.

Phase 3: Clean before you import

Migrating dirty data just relocates the problem, and a new system makes it harder to spot. Clean the export file, not the live CRM.

1. Deduplicate. Match on email first, then domain plus name. Our guide to CRM deduplication covers merge rules that preserve history.
2. Verify emails. Old databases carry 20–30% dead addresses. Run the file through verification so you do not import bounces — see our email verification guide.
3. Standardise formatting. Consistent country codes on phone numbers, title case on names, normalised company suffixes.
4. Fill critical gaps. Records missing company or job title are hard to segment later. Data enrichment is far cheaper before import than after.
5. Flag stale records. Anything with no activity in 24 months goes to an archive segment, not the main database.

Phase 4: Run a pilot import

Never migrate everything at once. Take a sample of 100–200 records that includes your most complicated cases: a contact with multiple deals, an account with several contacts, a closed-won deal with attachments, a record with unusual characters in the name.

Then check, manually:

• Did contacts stay linked to the right companies?
• Did deals keep their stage, value, close date and owner?
• Did notes and activity history come across with original timestamps?
• Did custom fields land in the right place with readable values?
• Did any automation fire on import and email a customer by accident? Turn off workflows before importing. This is the mistake that costs the most goodwill.

Phase 5: Full migration and cutover

1. Freeze the old CRM. Pick a window — typically a Friday evening — and tell the team no edits after that point.
2. Take a final export and a full backup. Keep the backup for at least six months.
3. Import in dependency order: users, then companies, then contacts, then deals, then activities. Reversing this order is what breaks relationships.
4. Reconnect integrations one at a time and test each before enabling the next.
5. Re-enable automation last, once you have confirmed the data is correct.

Phase 6: Validate before you declare victory

Return to the baseline counts from Phase 1 and compare. Any variance needs an explanation — intentional deletions are fine, unexplained gaps are not.

• Record counts by object type, within your expected tolerance.
• Total open pipeline value in the new CRM matches the old one.
• Every deal has an owner and a stage.
• Spot-check twenty random records against the source export.
• Run your three most important reports in both systems and compare the numbers.

The pre-cutover checklist

Print this. Do not start the full import until every line is ticked.

1. Baseline record counts documented
2. Full backup of source CRM saved
3. Field mapping spreadsheet signed off
4. Picklist collapse rules decided
5. All users created in the destination CRM
6. Duplicates merged in the export file
7. Email addresses verified
8. Stale records archived separately
9. Pilot import of 100+ records reviewed manually
10. Automation and workflows disabled
11. Integrations inventoried with a reconnection order
12. Edit freeze communicated to the team
13. Rollback plan written down
14. Owner assigned for post-migration validation
15. Training session scheduled for the week after cutover

After the migration: keep it clean

A migration is a rare chance to reset your data standards. Use it. Agree on required fields at record creation, set a naming convention for companies, assign one person to own database quality, and schedule a quarterly cleanup. Otherwise the new CRM degrades on exactly the same curve as the old one — our piece on CRM data decay explains how fast that happens.

Migrations are judged months later, on whether the team trusts the numbers. Slow down in phases one to three and the rest is mostly mechanical. For ongoing hygiene once you are live, see our CRM management guide.