Environment copies, restores and journeys asset ALM

What it does

Covers what actually happens to Customer Insights - Journeys when a Dataverse environment is copied or restored, and how digital assets and emails move between environments through solutions. Journeys is only partly a Dataverse application — a copy brings the records but not the service connection, the images or the interaction history.

Key facts

Copy and restore only work on Sandbox environments. Convert a production environment to sandbox first if you need to restore into it, and check the target isn't left in administration mode afterwards.
Interaction records (opens, clicks, website visits) and marketing service files (images used in emails and pages) are not copied. The exception is restoring a backup onto the same environment, where analytics and interaction data are kept.
Everything that was live comes back in Draft on the target and has to be published again.
Journeys behave differently from other records: draft and expired journeys are left alone, others are cloned in Draft on the target while the originals are set to Expired on the source. After a restore, journeys sit in Stopped and cannot be restarted directly — duplicate the journey, then run the copy.
The target environment must be reinstalled from Power Platform admin centre > Dynamics 365 apps after the copy. Without that step the environment has the solutions but no services.
Domain authentication does not survive. You re-authenticate every domain on the target and republish DNS records before it can send.
The target environment needs its own Journeys licence. If none is available, the copy finishes in a disconnected state and the key features simply don't work until you apply one.
Digital assets move between environments through managed solutions only. Add existing files from the digital assets library to a solution in Power Apps, publish, export managed, import on the target — URLs then resolve to the target environment.
Files and emails cannot go in the same solution, and adding an email does not pull in the digital assets it references. Import the digital assets solution first, then the emails solution; do it the other way round and every image link points back at the source environment. Files can't be removed from a managed solution afterwards either — deleting a digital asset means deleting the whole solution.

When to use / skip

You'll copy production down to a sandbox on almost every project — for testing an upgrade, reproducing an issue, or giving a partner a realistic environment. Do it, but plan the post-copy work rather than treating it as a Power Platform operation the client's admin can run unattended.

Restoring a backup onto a production environment carrying live journeys is a different matter. Consent state rolls back with the database, which is a compliance problem rather than a data one, and journeys come back stopped. Reserve it for genuine disaster recovery, and get the client to accept in writing that consent will revert.

For moving content — emails, images — between environments, solutions work well enough for digital assets and emails. Don't expect them to be a general-purpose journeys deployment pipeline; the ALM story here is narrower than it is for ordinary Dataverse customisation.

Configuration decisions

Copy type: "Everything" or "Customizations and schemas only". Schema-only avoids the risk of a sandbox with real contact data in it, at the cost of realism.
Whether the sandbox is allowed to send email at all, and how you prevent it — a sandbox with production contacts and a working domain is a live risk.
Whether interaction and analytics history needs preserving before a copy to a different environment, and if so exporting it through the Fabric integration first.
Which digital assets are packaged into a managed solution as part of the release, and who owns that solution.
The solution import order for any release that touches both assets and emails, written into the deployment runbook rather than remembered.
Whether old DNS records from the source environment's domain authentication are removed when a target is repurposed, being careful not to remove records another environment still uses.

Gotchas

The reinstall step after a copy is the one people miss. The environment looks fine, records are all there, and nothing sends. Symptoms point everywhere except the missing service connection.
Consent data reverting with a restore is the sharpest edge in this whole area. Export consent records to Excel before a restore so you have a reference for what changed.
Republished emails keep referencing image URLs from the source environment. They may even render, if the source still exists, which makes the problem invisible until the source is decommissioned.
Journeys restored into Stopped cannot simply be started. Duplicating them is the supported route, and the duplicates carry new IDs, which breaks any reporting keyed on journey ID.
Emails imported without their digital assets solution have every link pointing at the source. The documented fix is to bump the emails solution version and re-import — after the assets are in place.
Republishing after a restore is what stops journeys re-running and double-messaging people. On a same-environment restore, periods already elapsed won't re-fire, but don't rely on that across environments.

Consultant notes

Write the post-copy checklist once and keep it with the project: disable administration mode, reinstall the app, re-authenticate domains, republish assets and emails in that order, check the service account roles. Hand it to the client's admin team.
Make sure the client understands that a sandbox refresh is not free — someone spends half a day on DNS and republishing afterwards. Budget it in every release cycle that includes one.
Push for a "no sending from sandbox" rule enforced by not authenticating a real domain there, rather than by hoping nobody publishes a journey.
Demo the digital assets solution round trip once, in front of the client's ALM owner. The files-before-emails ordering is unintuitive and needs to be in their runbook, not just in yours.
Before any restore on production, take an export of consent records and agree with the privacy owner what happens to consent captured after the backup point.

Worth another look if Microsoft broadens solution support for journeys and segments, or changes what a copy preserves.