Methodology

Every event above is a period in which our checks of Microsoft 365 failed from two or more distant regions, meaning regions on different continents (see how we tell the difference). One that nothing else backs stops counting as an outage after 6 hours, and an event that fails these rules is retracted, kept, and listed on the corrections page. The time our checks failed is the wall-clock time inside the window an event's regions were failing; regions failing together still count as one event, not one per region. It is not proof of downtime for your account.

These figures come from our own checks of one public endpoint per service. They are not the service's real uptime.

Microsoft 365's claim at the time comes from their own public status feed, polled on our own schedule, and is shown separately from our check readings. We never change our own reading based on what a vendor claims, and their claim never becomes our reading.

This is one external reading from RealUptime's own regions, not Microsoft 365's own SLA measurement, which is usually defined against their own instrumentation and may cover a different scope than what we check. The credit estimate above is an estimate from our checks, and Microsoft 365 decides what counts. It is a starting point for a conversation with Microsoft 365, not a legal determination, and RealUptime is not counsel.