Website Migration Go/No-Go Checklist: Decide When to Switch or Roll Back
How should a small business decide whether a migrated website is ready to go live?
Use a written go/no-go card immediately before the switch. Confirm that the destination is reachable in its intended form, priority pages and customer journeys work, required integrations have been checked, changed URLs have an agreed destination, monitoring and decision owners are available, and a usable recovery option remains. Mark each item pass, accepted exception or stop. Do not proceed when a failed check could prevent customers from completing an essential task or leave the business unable to recover.
Define what can stop the launch
Makers’ Den says CMS switches, domain moves, HTTPS changes and redesigns can carry search and user-experience risks. Define stop conditions before launch: a failed essential customer task, unresolved routing, unavailable recovery option or missing decision owner should pause the switch. Allow a limited exception only when a named owner understands its consequence, accepts it and has assigned the fix.
Sources: The Definitive Guide on Website Migrations - Makers’ Den.
- Stop for failed essential customer tasks.
- Stop for unresolved routing or unavailable recovery.
- Do not confuse visual polish with operational readiness.
Run the customer-journey checks
Test the customer journeys that matter to the business, such as finding a service, contacting the business, submitting a form and receiving the expected confirmation. Check priority pages, navigation, calls to action and every connected service that supports an essential task.
Use website accessibility testing for important customer tasks, and apply privacy and consent checks to analytics, advertising or chat tools across relevant visitor states.
- Test priority pages and navigation.
- Test forms, calls to action and confirmations.
- Test connected services needed for essential tasks.
Check routing and changed URLs
Makers’ Den says major changes to URL structure, hosting, platform, design or content can affect how search engines crawl and index pages. Check the intended domain, secure page loading, priority URLs and the agreed destination of every changed priority address. Record an unexpected result as a stop or accepted exception rather than assuming it will resolve after launch.
Sources: The Definitive Guide on Website Migrations - Makers’ Den.
- Confirm the intended domain.
- Check secure loading and priority URLs.
- Check the agreed destinations for changed addresses.
Confirm owners, monitoring and recovery
Name the person who approves the cutover, the person making the technical change, the person checking customer tasks and the person responding to issues. Confirm that a current backup or previous arrangement can support recovery before switching.
Use the guidance on how to manage recurring website tasks to assign post-launch checks, fixes and continuing care.
- Name the final approver.
- Name technical, customer-check and response owners.
- Confirm a usable recovery option.
Record pass, exception or no-go
Record every check as pass, accepted exception or stop, together with the observed result, consequence, action owner and next action. Record the final approver and recovery decision so the team knows whether to proceed, pause or restore the previous arrangement. Before a service-affecting or irreversible step, confirm the provider’s current prerequisites in its own documentation.
- Mark pass, accepted exception or stop.
- Record consequence and assigned action.
- Record the approver and recovery decision.
Launch-day go/no-go and rollback decision card
Use this card immediately before the switch. Mark every item pass, accepted exception or stop; a stop condition means pause and use the recorded recovery decision.
| Check | Decision | Observed result or consequence | Owner, next action and approval |
|---|---|---|---|
| Destination reachable in intended form | Pass / accepted exception / stop | Record the observed result and consequence | Record owner and next action |
| Priority customer journeys | Pass / accepted exception / stop | Record the observed result and consequence | Record owner and next action |
| Required integrations | Pass / accepted exception / stop | Record the observed result and consequence | Record owner and next action |
| Changed URLs have agreed destinations | Pass / accepted exception / stop | Record the observed result and consequence | Record owner and next action |
| Decision, monitoring and response owners available | Pass / accepted exception / stop | Record the observed result and consequence | Record owner and next action |
| Usable recovery option remains | Pass / accepted exception / stop | Record the observed result and consequence | Record owner and next action |
| Final cutover decision | Pass / accepted exception / stop | Record the recovery decision | Record final approver and next action |
This card supports a business decision and does not eliminate technical, customer-experience or search risk. Confirm current provider prerequisites before an irreversible or service-affecting change.
When this is not the right approach: This card is not enough on its own for a complex system whose critical functions cannot be checked by the business team. Use specialist technical support when the required checks or recovery option cannot be understood and verified.
Related guidance
What follow-up questions matter most?
- What should trigger a no-go decision?
- Stop when a failed check could prevent customers completing an essential task, routing is unresolved, a recovery option is unavailable or nobody can approve and manage the response. Record the issue rather than treating it as a minor cosmetic defect.
- Can any defects be accepted at launch?
- A limited defect may be accepted only when its consequence is understood, a named owner accepts it and a fix is assigned. Essential customer tasks and recovery controls should not be treated as casual exceptions.
- How should the final decision be recorded?
- Record each item as pass, accepted exception or stop, together with the observed result, consequence, owner, final approver and recovery decision. This makes the decision reviewable instead of relying on memory.
What steps does this workflow follow?
Run a website cutover decision card
- Define stop conditions: Agree which failures prevent launch before the final technical switch is attempted.
- Check customer tasks: Test priority pages, navigation, forms, calls to action, confirmations and relevant connected services.
- Check routing: Confirm the intended domain, secure loading, priority URLs and agreed destinations for changed addresses.
- Confirm response ownership: Name the final approver, technical operator, customer-check owner and issue-response owner, with a usable recovery option.
- Record the decision: Mark each item pass, accepted exception or stop, then record the approver and recovery decision.