Desk with laptop, emergency checklist, phone and local folder illustrating a tested cloud contingency plan

Microsoft 365 outage: Can your team complete the continuity drill?

The file is in a second cloud. The proposal still cannot go out. A 30-minute drill shows where your continuity plan actually breaks.

Short answer

How do you test a Microsoft 365 continuity plan?

Ask a second person to complete a real task with sample data: sign in from scratch, edit a current file and send the result. If any step still depends on the unavailable service, the fallback is not yet independent.

Imagine a Monday morning. A proposal has to go out by noon. There is a copy of the file in a second cloud, but the team is still waiting. The colleague with access is away; the sign-in code for the backup account is sent to the company mailbox that is unavailable. The file exists. The route to getting work done does not.

This is a hypothetical drill, not a customer incident. It exposes the gap between “we have a backup” and “we can keep working”. That gap is worth testing.

What the Microsoft 365 outage showed – and what it did not

Several Microsoft 365 functions were affected on 12 and 13 November 2026. Impact varied by service and organisation; administrators should check Service health for their own tenant. The incident does not establish that everyone was affected in the same way, or that another provider is automatically independent.

The sequence matters for continuity planning: sign-in may work again while mail delivery or connected applications are still recovering. During the same incident, OpenAI and Anthropic reported problems with their Microsoft 365 connections. Their assistants did not have to be unavailable as a whole; a disrupted path to the required data was enough.

Plan B is a chain, not a second logo

For one important workflow, write down four questions: Who can sign in? Where is the current file? Which application can use it? How will the result reach its recipient? Put the service each answer depends on beside it. If Microsoft 365 appears more than once, that is not necessarily a mistake. It tells you which parts of the chain your drill must interrupt.

A filename in File Explorer, for example, is not proof that a file works offline; it may be a placeholder. Even a downloaded document may, depending on its protection settings, require a connection to its original environment. Test a representative file under your organisation's actual rules: download it fully, disconnect, open it, edit it and save it. The result depends on your Microsoft configuration.

Thirty minutes can tell you more than a backup report

Arrange a safe drill with your IT team outside a real outage. Use test accounts and sample data only. The task is for a second authorised person to update a sample proposal and deliver it to an external test recipient without using the Microsoft services you are treating as unavailable.

  1. Start from scratch. Use a fresh browser profile, not an existing session. Where do the password and second factor come from? Who helps if the usual owner is absent? Do not disable multi-factor authentication for the drill.
  2. Use a current file. Create a new test version shortly before the drill. Can the second person find, open and edit that exact version? Record how old the available copy is.
  3. Deliver the result. Can the external recipient open the correct version? Does the agreed return channel work? A file sitting in backup storage has not finished the job.
  4. Plan the return. Who merges changes from the fallback environment when normal service returns? Without an owner, two plausible versions can quietly diverge.

For each step, record pass or fail, the service and person needed, and the time to a usable result. Test local offline copies and a second online service separately. Disconnecting the network does not also prove that another provider is reachable independently.

How luckycloud can fit into the drill

Our Sync Client can keep selected folders on a computer. Confirm that syncing has finished before the drill. The Drive Client instead presents files as a virtual drive for online access; it is not a local copy. If your Enterprise sign-in uses Microsoft Entra ID, include that identity dependency in your test. Moving a file does not, by itself, make sign-in independent.

For the return channel, you can also set up luckycloud Mail & Calendar in advance and test it with a second person. A new mailbox does not automatically receive messages addressed to your existing Exchange mailbox. Our guide to emergency communication explains how to make a fallback address useful.

What matters in the end is not a list of subscriptions. It is a sentence your team can substantiate after the drill: “We finished the proposal without the usual service, and the recipient received it.” If you cannot say that yet, you at least know exactly where the plan needs work.

Sources and further reading

You might also be interested in