Choosing your outbound-to-real-time migration path

What it does

Outbound marketing is gone. Microsoft removed the module from all environments in May 2026, after a phased removal that started in July 2025. If you are still holding outbound assets, this is a data-archaeology and rebuild exercise, not a migration in any tooling sense.

Key facts

The outbound sitemap area, its services and its Insights reports were removed in May 2026. The underlying tables were not dropped — they are still reachable through the Power Apps maker portal and advanced find.
Outbound emails, content blocks, forms, marketing pages, customer journeys, templates, lead scoring models, social posts and subscription lists all still exist as rows. None of them can be used in real-time journeys. There is no converter.
The internal segment membership table was removed, so outbound segment records survive but their member lists do not. Treat outbound segment definitions as documentation, not data.
The asset library carries over intact — images and files uploaded under outbound are available in real-time journeys.
Events are a shared capability. Event tables and data remain and continue to work.
Marketing lists are a Sales entity, not an outbound one, and remain usable as an audience source in real-time journeys.
Outbound analytics data was merged with real-time journeys data. Existing custom Power BI reports break: both the format and the location changed, and you rebuild against the Microsoft Fabric data sources.
Consent is the one thing with a real migration tool. The Load consent button in the consent centre pulls donotemail, donotbulkemail and msgdpr_donottrack from contacts and leads, and maps outbound subscription lists onto topics under a purpose. It is one-way — there is no reverse migration.
Any custom UI, plugin or integration that wrote to outbound assets and depended on outbound services may fail. That failure is silent from the user's point of view until someone notices records aren't being created.

When to use / skip

There is no skip. The only live question is how much of the old estate you rebuild and how much you quietly retire, and the honest answer on most tenants is that you rebuild about a third of it.

Go through the outbound journey list and sort it into three piles. Pile one is the handful of journeys that actually ran in the last twelve months and matter commercially — nurture, event follow-up, the monthly newsletter. Rebuild those properly in real-time, and take the opportunity to redesign rather than transcribe, because segment-based recurring journeys behave differently from outbound's tile model and a literal port produces something worse than either. Pile two is journeys that ran once for a campaign in 2022 and were never switched off. Retire them and say nothing. Pile three is the awkward middle — things a business unit swears it needs but cannot show you a send from. Make them justify it before you rebuild it.

Phase consent first, always. Everything else in real-time journeys is gated on having a compliance profile with sensible purposes and a populated set of contact point consent records. Run Load consent early, in a sandbox, and check what it produced before anyone writes an email. Then rebuild the core channel setup — domain authentication, brand profiles, quiet times, frequency capping — then content, then journeys. Analytics last, because the Fabric rebuild is a separate skill and a separate conversation with whoever owns the client's BI.

If a client is still on outbound in 2026, the more useful advice is usually "your sending domain reputation, your consent data and your contact hygiene are the real assets — the journey definitions are not". Frame the work that way and the scope argument gets easier.

Configuration decisions

Whether to rebuild in the existing environment or stand up a clean one. Existing is nearly always right — consent, contacts, events and asset library all live there — but a genuinely rotten configuration is worth escaping.
How consent maps. Outbound subscription lists become topics under a purpose, and you choose the purpose structure before you press Load consent. Getting the purpose wrong is expensive because purposes cannot be deleted, only made inactive.
Whether to run Load consent or import consent from Excel. If the client stored opt-out anywhere other than "Allow bulk email" — a custom two-option field, a preference table — Excel import is the supported route.
What happens to lead scoring. Real-time journeys has its own scoring models and they are not the outbound ones. Decide whether scoring is in phase one or deferred.

Gotchas

Consent migration collapses shared email addresses to the most restrictive answer. If several contacts share info@, the result is opted in only when every one of them allows bulk email — otherwise everyone on that address is opted out. On B2B data this can wipe out a chunk of a list.
Only one email address per contact is loaded, whichever one audience configuration names. Secondary addresses carry no consent.
Business units are consent-segregated. You load consent separately into each business unit's compliance profile, and forgetting one produces a business unit that silently sends nothing.
Restoring the environment reverts consent records to the backup state, which can leave obsolete opt-ins live. Export consent to Excel before any restore.

Consultant notes

Sell this as a rebuild with a consent migration attached, not a lift and shift. Clients who have been told "there's a migration tool" are thinking of Load consent and nothing else, and that expectation needs correcting in the first workshop.
Demo the consent centre after a sandbox Load consent run. Seeing real contact point consent records is what makes the scope land, and it exposes shared-address damage before it matters.
Push back hard on porting journeys one-for-one. Ask for the last twelve months of sends per journey; the list shortens itself.
Check domain authentication, DKIM and the sending warm-up plan before the first real send. Reputation built under outbound follows the domain, not the module, but a long gap in sending undoes it.
Before go-live, run the maker portal over the solution for anything still referencing outbound tables — plugins, flows, canvas apps, integration endpoints.

Worth another look if Microsoft changes what remains queryable in the outbound tables, or if the archived documentation PDF moves.