Move information without losing meaning or control
A successful migration preserves the information the organisation needs, its context, quality and appropriate access. Copying every field is not success if records become incomplete, duplicated, misclassified or impossible to reconcile.
Inventory sources, owners, fields, volumes, quality, retention and dependencies. Decide what should move, archive or be deleted, map and transform it, protect a recoverable source copy, test representative and difficult records, reconcile totals and samples, then use a controlled cutover with a rollback rule.
Use this migration sequence
- Discover: identify data stores, owners, meaning, formats, volumes, quality, access, integrations and record obligations.
- Decide: classify what moves, stays available temporarily, is archived, corrected, merged or securely deleted.
- Map: define source-to-target fields, identifiers, validation, transformations, defaults, code values and rejected-record handling.
- Protect: restrict migration access, create recoverable copies, log movement and use safe representative test data.
- Test: migrate normal, edge, historical, incomplete and high-consequence records; test downstream reports and integrations.
- Reconcile: compare record counts, totals, relationships, exceptions and sampled meaning, with owners signing off.
- Cut over: freeze or control changes, run the agreed sequence, communicate status and invoke rollback if thresholds fail.
- Close: resolve exceptions, retain required evidence, remove temporary access and copies, and retire the source under policy.
Define acceptance before the first test
| Check | Example evidence | Owner |
|---|---|---|
| Completeness | Expected counts by record type, period and business unit match or have approved exceptions. | Information owner |
| Accuracy and meaning | Sampled values, relationships, dates, units, codes and attachments remain correct in context. | Process owner and users |
| Operation | Priority workflows, searches, reports, integrations, retention and permissions work with migrated data. | Service owner |
| Recoverability | Rollback point, source preservation and restoration procedure have been tested to the agreed level. | Technical owner |
Minimise before you migrate
Do not move personal or sensitive data merely because it exists. Confirm purpose, accuracy, retention, legal basis and access with the responsible specialists. The UK ICO’s data-minimisation guidance says organisations subject to UK law should hold adequate, relevant data limited to what is necessary. Requirements differ elsewhere.
GOV.UK guidance on managing legacy technology recommends an accurate data-asset register, while its migration guidance stresses stakeholder agreement and a controlled changeover.
Business data migration FAQs
Should a business migrate all historical data?
Not automatically. Keep what the organisation needs for a defined operational, legal or evidential purpose, in an appropriate system and format. Archive, delete or anonymise only under the relevant policy and jurisdiction-specific requirements.
How can a team prove that migration succeeded?
Use agreed reconciliation measures such as counts, totals, relationships, exceptions and sampled record meaning, then test real workflows, permissions, reports and integrations. Record the evidence and the named owner who accepted it.
When is rollback needed during data migration?
Define rollback before cutover using thresholds such as failed reconciliation, unavailable critical workflows, unacceptable corruption, security failure or a missed time window. Also define who can invoke it and how changes made during cutover will be preserved.
Continue your business technology decision
Use the next guide that matches the requirement, investment, supplier, implementation, migration, access, continuity or renewal question you still need to resolve.
