Most CRM migrations that go wrong fail before any data moves: an object nobody listed, a picklist nobody mapped, an export that left the emails behind. This checklist puts the decisions in order. Each stage has a short list of checks, and where a native tool has a documented limit that changes the plan, it is called out.
What should a CRM migration plan include?
| Stage | Output | Owner |
|---|---|---|
| 1. Audit the source | Inventory of objects, record counts, fields, users, automations | CRM admin |
| 2. Decide scope | What moves, what’s archived, what’s dropped | Sales ops plus leadership |
| 3. Map | Object, field, picklist and owner mapping, signed off | CRM admin plus migration lead |
| 4. Build the destination | Properties, pipelines, users, labels, permissions | Destination admin or partner |
| 5. Back up and test | Full backup, sample run, sign-off | Migration lead |
| 6. Migrate | All objects loaded in order, with history | Migration lead |
| 7. Cut over | Delta of late changes, source set read-only | Migration lead plus admins |
| 8. Validate and rebuild | Reconciliation report, automation and reports rebuilt | Admin plus process owners |
Stage 1: What should I audit in the old CRM?
- Every object in use, standard and custom, with record counts.
- Every field per object, with fill rate. Fields under a few percent filled are candidates to drop.
- Picklist and dropdown values, including retired ones still on old records.
- Relationships: which objects link to which, and any many-to-many links or relationship types.
- Activity history: emails, calls, meetings, tasks and notes, with counts per type.
- Files and attachments, with total size.
- Users and owners, including inactive users who still own records.
- Automations, reports and integrations that touch the CRM.
- What the export leaves out. Some exports skip whole data types. Zendesk Sell’s export, for example, can’t include emails, call logs or activity history.
Stage 2: How do I decide what to migrate?
- Keep everything open: live deals, open tickets, active accounts.
- Keep history customers or auditors may ask about.
- Keep data your reports depend on, such as closed-won history for forecasting.
- Archive the rest in a searchable backup instead of deleting it.
- Agree the duplicate rules: which field identifies a duplicate, which record wins.
- Write the scope down and get it signed off.
Stage 3: How do I map objects and fields?
- Map objects first. Data models differ: Salesforce has separate leads and record types; HubSpot uses lifecycle stages and pipelines; Zoho has modules and sub-forms.
- Map fields with types. A text field can’t land in a number field. Dates need one format.
- Map picklist values one by one, including stages and statuses.
- Map owners by email to users in the destination.
- Decide where original dates go. HubSpot sets the contact Create date automatically, so carry the source date into a custom property. Salesforce lets API tools set Created Date and Created By once an admin enables Create Audit Fields.
- Flag fields the destination can’t create on the fly. Zoho’s Migration Wizard, for example, can’t create Formula or Autonumber fields and creates at most 50 new fields during migration.
Stage 4: What do I set up in the new CRM first?
- Users, roles, teams and permissions.
- Custom fields and properties, with dropdown values matching the mapping.
- Pipelines and stages.
- Custom objects and their associations.
- Association labels or relationship types. HubSpot needs custom labels to exist before import, and they require Professional or Enterprise.
- Duplicate rules and validation rules, switched off for the load if they would reject historical data.
- Automations left off until the data is in, so old records don’t trigger emails or tasks.
Stage 5: How do I back up and test?
- Take a full backup of the source. In Salesforce, the Data Export Service includes attachments if you select them; its zip files are deleted 48 hours after the email, so download them straight away.
- Run a sample that covers every object, including custom objects, records with many activities and records with attachments.
- Check samples in the UI, not just counts: open ten records and compare them field by field with the source.
- Have users sign off the sample before the full run.
Stage 6: What order do I migrate in?
- Users (or owner mapping).
- Companies or accounts.
- Contacts and leads.
- Deals or opportunities, with line items.
- Tickets or cases.
- Custom objects, parents before children.
- Activities: emails, calls, meetings, tasks, notes, each linked to its parent records.
- Files and attachments.
Loading out of order is the most common cause of orphaned records: a contact loaded before its company has nothing to attach to.
Stage 7: How do I cut over with minimal downtime?
- Pick a cutover window and tell users.
- Capture records created or changed in the source since the main run, and migrate that delta.
- Set the old CRM to read-only for users.
- Repoint integrations, forms and email sync to the new CRM.
- Switch automations on in the new CRM.
Stage 8: How do I validate the migration?
- Compare record counts per object, source against destination.
- Compare totals that matter: pipeline value, closed-won by year, open ticket count.
- Spot-check records with the most activities and attachments.
- Check owners, create dates and stage values on a random sample.
- Rebuild workflows, reports and dashboards, and test them with real records.
- Keep the old system read-only, or keep the backup, until finance and sales sign off.
What are CRM migration best practices?
- Freeze the schema, not the users. Lock field and pipeline changes during the project, but use a delta to keep users working.
- Migrate history with the record, not as a later phase that never happens.
- Test on the hardest records, not the cleanest.
- Make it repeatable. You’ll run the migration more than once. A manual process that can’t be rerun is a risk.
- Keep a way back. Know how you would undo the load before you run it.
When to bring in FlitStack AI
FlitStack AI is a data-migration platform sold as software plus a done-for-you service. It covers stages 3 to 8 for the data: AI-proposed mappings you review, a sample run, migration in dependency order with timestamps and owners kept, delta migration for cutover, validation on every record and one-click rollback. The CRM implementation service handles the destination build and automation rebuild.
Estimating budget? Read what drives CRM migration cost. Browse routes on the CRM migration hub, or book a free scoping call.