From reading to action
A practical path through this guide.
Select a step to explore what to do and why it matters.
Inventory workflows and state
Include schedules, dependencies, credentials, checkpoints and pending exceptions.
Migrate an automation platform by preserving business behavior, identifiers, and recovery state. Rebuilding a diagram in another tool does not establish equivalence. A safe plan accounts for records arriving during the transition and for actions already completed by the old system.
Inventory before rebuilding
List active workflows, schedules, webhook destinations, credentials, external dependencies, stored state, and exception queues. Identify undocumented workflows still running under personal accounts. Separate obsolete flows from ones the business still needs.
For every active workflow, collect representative sanitized fixtures and expected destination outcomes. Include edge cases such as duplicate events, missing fields, and a temporary outage. These become portable acceptance checks across the two platforms.
Move state deliberately
Determine which deduplication keys, checkpoints, and pending approvals must survive. A new platform with an empty deduplication store may treat replayed events as new business. A reset pagination checkpoint may repeat an entire historical import.
Translate field types and time handling carefully. Test null values, arrays, decimal amounts, timestamps, and character encoding. A transformation that “looks the same” in an editor can still serialize data differently at the API boundary.
Plan the cutover window
Use shadow processing or a controlled comparison where feasible, while preventing the replacement from performing duplicate external writes. Compare proposed destination changes before enabling them. Keep production credentials and side effects separate from a test copy.
Choose a cutover sequence that captures events arriving between the old system stopping and the new one starting. Depending on the provider, this may require a temporary intake queue or a reconciliation pull. Document how to roll back without replaying completed actions.
Verify and retire
- Compare eligible source records across the transition window.
- Check destination identifiers and important mapped fields.
- Resolve old exception queues or transfer their ownership.
- Update webhook registrations and alert destinations.
- Remove unused credentials and subscriptions after acceptance.
Keep the old configuration export and recovery notes for an agreed retention period. Record the final checkpoint and the time at which ownership changed. A migration is complete when the business outcomes reconcile and the team can operate the replacement, not merely when the old tool is turned off.
Further reading
See idempotent workflow design and handoff documentation for two common migration gaps.
A few practical answers
Frequently asked questions
What state needs to survive an automation platform migration?
Preserve deduplication keys, checkpoints, pending approvals and exception ownership where required. Losing that state can repeat old actions or skip unfinished records.
How can a replacement workflow be compared without duplicating writes?
Use controlled shadow processing or compare proposed changes while suppressing duplicate external actions. Define cutover and rollback responsibilities before moving live intake.
