When someone leaves, their mailbox still has mail coming to it. Deleting the account bounces those senders; leaving the account licensed and unmonitored means nobody reads them. Neither outcome is visible until a customer asks why they were ignored.
The pattern below handles it with three objects and no new habits. It preserves the departed employee’s address, routes anything sent to it at people who are already watching their own inbox, and adds no mailbox for anyone to remember to check.
The flow
What each piece is doing
The shared mailbox exists to keep the address alive. When you convert the user mailbox rather than deleting it, the address keeps resolving, the history stays put, and external senders never get a bounce. A shared mailbox under 50 GB doesn’t need a license of its own, so the address can stay alive without paying to keep a seat active indefinitely.
The distribution group exists so the recipient list is a config change instead of a mailbox change. When the person covering the role moves on in six months, you edit group membership — you don’t go back into mail flow settings on a mailbox you’d forgotten existed. It also scales: one recipient or five, the mailbox side never changes.
The recipients get the message in their own inbox, where they already work. That is the whole argument against the alternative. You can grant a few people Full Access to the shared mailbox and let it auto-map into Outlook, but then you’ve given them a second inbox that nobody has agreed to own, and mail sitting unread in a shared mailbox is indistinguishable from mail nobody received.
The steps
- Convert the departed user’s mailbox to a shared mailbox.
- Create (or reuse) a distribution group containing the people responsible for that person’s work.
- On the shared mailbox, set forwarding to that distribution group.
- Send a test message to the old address and confirm it lands with every group member.
Step 4 is not padding. Forwarding can be saved on the mailbox and still not produce the result the client expected — a group whose membership is wrong, or delivery restrictions on the group that drop the message. The only confirmation that counts is watching a test message come out the other end at every recipient.
The one setting to confirm
Forwarding on a mailbox comes with a companion option — keep a copy of forwarded messages in this mailbox. It’s a single checkbox, and it decides whether the shared mailbox stays a record or becomes a pass-through.
Confirm the choice with the client rather than picking a default:
- Keep a copy if anyone might need to search that address later. The mailbox stays a complete record of everything sent to the departed employee, independent of what the recipients did with their copies.
- Don’t keep a copy if the retention policy says the mailbox should empty out, or if the address is meant to be a forwarder and nothing more.
There’s no universally right answer — it depends on the client’s retention, auditing, storage, and operational requirements. What matters is that the choice is made deliberately and recorded on the ticket, not defaulted to whatever the checkbox happened to be set to.
One thing that makes this pattern simpler than it looks: the forward target is a distribution group inside the tenant, so it’s internal delivery. The outbound anti-spam controls that block or flag automatic forwarding to external addresses aren’t in play here. A request to forward straight to someone’s personal address is a different question with a different answer.
The failure mode to avoid
The version of this that ages badly is forwarding pointed at an individual instead of a group. It works on day one, and it keeps working right up until that individual changes roles or leaves — at which point the fix means going back into mail flow settings on a mailbox nobody remembers exists, assuming anyone notices at all. Chain two of those together and the mail is being forwarded through a mailbox that is itself forwarding somewhere else.
Sending to a group costs nothing extra at setup and makes every future change a membership edit. That is the entire reason the middle box is there.
Four steps, one checkbox, one test message. Then close the ticket.