Contacts are rarely what goes wrong
Contacts are usually the easiest part of a migration. The trouble comes in everything attached to them. Here are the risks in the order they tend to appear, with the check that catches each one.
| Risk | How it shows up | Check that catches it |
|---|---|---|
| History not exportable | The source’s CSV export has no emails, calls or notes. Zendesk Sell’s own export, for example, can’t include activity history, emails or call logs | Read the source export from its API, not from a CSV |
| Broken associations | Deals load without their company; activities without their contact | Load parents before children and count orphans after the sample |
| Reset dates | Every record shows the import date as its created date | Map the original created date to a field you can write, and test it |
| Unmapped owners | Records land unassigned or assigned to the person who ran the import | Create the users first and match them by email |
| Duplicates | A rerun or a second source doubles the records | Store a source ID on every record and update on it |
| Automations that don’t move | Lead routing, alerts and sequences stop on day one | List them in the audit and rebuild them before cutover |
| Changes you can’t reverse | A bad import overwrites good data | Export the destination before every run |
What native tools do / where they stop
Native importers are built for loading data, not for spotting these risks. HubSpot’s Smart Transfer creates properties and pipelines for supported apps, but it doesn’t move custom objects. HubSpot’s importer can delete the records it created, but it can’t reverse updates to existing ones. No native importer checks for orphaned activities or compares what landed against the source.
How to reduce the risk
- Pull the source data through its API, and treat the CSV export as incomplete.
- Run a sample that covers every object and activity type.
- Compare the sample against the source field by field, including dates and owners.
- Snapshot the destination, then run the full load.
- Use a delta to catch changes made during cutover.
- Rebuild automations and test them on the migrated data before users switch over.
Where FlitStack AI fits
FlitStack AI pulls data through each platform’s API, preserves timestamps, matches owners, checks every record with hash-based validation, and rolls back in one click. The service team can also rebuild processes in the destination. Start with the CRM migration checklist.