A regional organization still ran Exchange on-premises with hybrid leftovers from an earlier partial move. Patching risk was rising, high availability was thin, and nobody wanted the classic Monday morning: mailboxes online, but scanners, ERP alerts, and line-of-business apps silently failing to send.
- On-prem → OnlineMailbox batches completed
- Full inventoryApps, scanners, relays
- Relay redesignBefore final decommission
- Hybrid cleanupDocumented retirement path
What success meant here: user mailboxes in Exchange Online, device and application mail paths revalidated, and a clear answer to “what still needs an Exchange server, and for how long?”
The situation
Some mailboxes had already wandered into the cloud in prior years. Others remained on-prem because of public folders, legacy apps, or simple inertia. Hybrid made the environment feel “modern enough” until certificate renewals, cumulative updates, and weekend patch windows started dominating IT’s calendar.
The quiet risk wasn’t Outlook, it was everything else that sends mail: multi-functions, monitoring systems, ERP notifications, and a few scripts nobody had documented since 2019.
What discovery actually found
- Mailboxes and archives still on Exchange Server, including a few shared and resource mailboxes people forgot existed
- Public folder usage that was smaller than folklore claimed, but not zero
- SMTP clients pointing at on-prem receive connectors with inconsistent authentication
- Transport rules and journaling assumptions that needed a cloud equivalent
- Hybrid components still required for some management paths even after most users could live online
How we sequenced the work
1. Inventory before migration batches
We listed every mailbox, connector, device, and application path. That list became the project backbone. If it wasn’t on the list, it wasn’t “out of scope”, it was a future incident.
2. Hybrid readiness and batch design
Batches were sized for support reality, not maximum theoretical throughput. Early batches included IT and a friendly department. Later batches moved denser operational groups once the pattern was boring.
3. SMTP relay redesign in parallel
We didn’t wait until the last mailbox moved to care about scanners. Relay options were designed and tested while batches ran, so cutover week wasn’t a scavenger hunt for which copier still pointed at an old IP.
4. DNS and mail-flow validation
MX and related records moved with a checklist and rollback notes. Validation included ordinary user send/receive and the application alerts that only fire when something’s already wrong.
5. Hybrid residual cleanup
After user mail was stable online, we documented what hybrid pieces could shrink, what still had a temporary reason to exist, and what the retirement path looked like for the remaining on-prem role. Operations got a written story, not tribal knowledge.
Migration vs. relay workstreams
| Workstream | Goal | Failure mode we avoided |
|---|---|---|
| Mailbox batches | Users on Exchange Online | Big-bang weekend with no pilot learning |
| App/device SMTP | Reliable outbound notifications | “Mail works” while ERP alerts die |
| DNS / flow | Clean authoritative path | Split-brain routing surprises |
| Hybrid cleanup | Less on-prem to patch | Forever-hybrid with no owner |
Results
- User mailboxes moved to Exchange Online through staged batches with predictable support load
- Device and application mail paths revalidated, no surprise outage for scanners or ERP alerts
- On-prem Exchange footprint reduced to what was still justified, with a documented next step to retirement
- IT spent fewer weekends on server patch theater and more time on identity and client experience
The app-relay inventory alone saved us from a Monday morning fire drill.
- Messaging & Collaboration Admin
The lesson we keep repeating
Mailbox migration is the visible half. The other half is everything that sends mail without a human at a keyboard. If you only migrate people, the apps will write the postmortem for you.
Still running Exchange on-prem or messy hybrid?
We’ll inventory mailboxes, relays, and hybrid dependencies, and give you a staged plan with clear decommission criteria.
Book a free consultation