Skip to content
PMInterviewKit

The outage question in a PM interview: a five-step answer

“Your product is down for millions of users. You are the PM. What do you do?” Interviewers ask this in many forms. The Microsoft 365 outage of November 2026 is a live example, so this page uses it to show how a strong answer goes.

Last updated 13 Nov 2026 · By PMInterviewKit · Not affiliated with Microsoft

Ava walks through the five steps.
Read the transcript

Outlook and Teams went down on Thursday. Now picture the interview question: your product is down for millions of users. You are the PM. What do you do? I'm Ava. Here are five steps. First, what we know. Microsoft said some of its servers became unavailable, and the rest took the extra load. It said this was not a security breach, and that the root cause will come later, in a post-incident report. So we will not guess. We will answer the question. Step one. Size the impact. Who is affected, how many, and how badly? Step two. Stop the damage first. Roll back, fail over, or switch to a fallback. Engineers lead that work. Root cause and lasting fix come later. Step three. Tell people early, on a steady rhythm. Say what you know, what you do not know, and when the next update comes. Microsoft kept posting updates. Step four. Make the hard calls. Decide what to protect and what to pause, and say what you are giving up. Step five. Afterwards, learn from it. Run a blameless review, prevent a repeat, and find a way to know sooner. Microsoft has said a post-incident report will follow. A PM does not repair servers. A PM sizes the impact, protects people, and keeps everyone informed. The full outline is free on our site.

What happened

On Thursday, November 12, 2026, Microsoft 365 services including Outlook, Teams and Copilot stopped working for many people. The trouble ran into Friday, November 13, and Microsoft was still working on a fix when this page was last updated.

The BBC reported that Downdetector saw problem reports from more than 70 countries. At around 18:00 GMT on Thursday, nearly 140,000 people had flagged problems with Outlook. More than 430 million people use Microsoft 365 apps, the BBC reported.

Microsoft said some of its servers became “unexpectedly unavailable”, which put a heavier load on the rest. It said it was not aware of any malicious activity and that this was not a security breach.

Its public updates were frank about uneven progress. Microsoft said it will give the root cause in a post-incident report once the problem is under control.

The BBC reported disruption at health services, pharmacies and police forces, so this was a hard day for many people. This page does not guess what went wrong or say how Microsoft should fix it. It uses the outage as a live example of an interview question.

The interview question

Your product is down for millions of users. You are the PM. What do you do?

You may hear it as “checkout went down for an hour” or “the app is crashing after a release”. The test is the same: calm, order and judgment under pressure.

A PM does not repair servers. A PM sizes the impact, decides what to protect, keeps people informed, and makes sure the team learns from it.

The five-step answer

  1. 1.Size the impact

    Start with who is affected, how many, and how badly. Is it every user or one region? Are people blocked, or only slowed down? Is anyone relying on the product for something urgent? Check your dashboards, the support queue and your largest customers.

  2. 2.Stop the damage first

    The first goal is to get people working again, not to explain the failure. That usually means a rollback, a failover to healthy systems, or a simpler fallback. Engineers lead that work. Your part is to answer product questions quickly and keep others from adding noise. The root cause can wait.

  3. 3.Tell people early, on a steady rhythm

    Customers, support teams and your own leaders should hear from you before they have to ask. Post to the status page, brief support, and update leadership. Each update says what you know, what you do not know yet, and when the next update will come.

    In this outage: Microsoft kept posting updates while it worked.

  4. 4.Make the hard calls

    This is the part only a PM can do. Decide what to protect first, such as customers for whom the product is urgent. Decide what to pause, such as a planned launch or a marketing email. Then say what each call gives up.

  5. 5.Afterwards, learn from it

    Once service is back, run a blameless review. Agree on the fix that prevents a repeat and on monitoring that would tell you sooner. Measure what the outage cost, and follow up with the customers who were hit.

    In this outage: Microsoft has said it will publish a post-incident report with the root cause.

The 60-second spoken version

“First I would size it: who is affected, how many, and how badly. Second, I would back the engineers in stopping the damage before anyone hunts for the cause: a rollback, a failover or a fallback. I would not debug it myself. Third, I would get an update out early to customers, support and leadership, and set a rhythm: what we know, what we do not know, and when the next update comes. Fourth, I would make the product calls: what we protect first, what we pause, and what that costs. Afterwards, I would run a blameless review, agree on the fix and better monitoring, and follow up with affected customers.”

Three mistakes candidates make

Jumping to the root cause

Many candidates start guessing at causes in the first sentence. In a real outage that adds noise while engineers work. In an interview it signals that you would get in the way. Stop the damage first.

Going silent

Waiting for the full story feels safe. People who hear nothing assume the worst, and support gets every question you did not answer. A short, honest update now beats a complete one in three hours.

Promising a fix time

You do not control when the fix lands, and a missed promise costs you trust a second time. Promise the time of your next update instead.

What to practice next

Source

Outage facts are Microsoft’s public statements and figures as reported by BBC News live coverage of the Microsoft outage, read 13 Nov 2026, about 09:30 GMT. Product names belong to their owners.