← All insights Guide

Why did my Power Automate flow just stop running?

Nine causes of a flow that quietly stopped, where to look for each in Power Automate, and what happens to the backlog when you switch it back on.

A flow that was working and stopped is, in the order I check, switched off or suspended, a broken connection, a lapsed licence, a trigger condition filtering every event out, a policy block or throttling. Most take under ten minutes to find in the flow's own page. Microsoft turns a flow off by itself after 14 days of continuous failure, and after 90 days with no trigger.

Work through it in this order

This follows Microsoft's own troubleshooting guidance, cheapest checks first. Open the flow in make.powerautomate.com, My flows, and start at the top of its details page.

#CauseWhere to lookFix
1The flow is off or suspendedStatus at the top right of the details pageOff: turn it on. Suspended: Power Automate stopped it after repeated failures, so read the run history for the failing action, fix that, then turn it on
2The trigger check was skipped28-day run history, All runsA skipped check means the trigger condition was not met. Read the condition expression; a common mistake is one that filters out every event
3A connection has brokenMore, then Connections, and the Status columnSelect the Fix connection link and sign in again. Passwords expire, and some organisations set the sign-in token to expire on a timer
4A licence has lapsed or was never enoughDetails, then Plan; the connector list shows PREMIUMPremium, on-premises and custom connectors need a standalone Power Automate or Power Apps licence
5A data loss prevention policy blocks itEdit and save the flow; the flow checker reports a violationYour admin has to change the policy. The message is "Looks like this workflow is disabled by your organization"
6The environment is in administration modePower Platform admin center, Environments, DetailsAdministration mode turns off every background process, including flows. Set it to Disabled
7The trigger cannot registerDetails page shows "There's a problem with the flow's trigger"Often a firewall allow-list missing Power Automate's addresses. Fix it, then save the flow, change it back, save again
8ThrottlingSlow or missing runs, HTTP 429, or a high action count in analyticsFewer actions, a trigger condition, or a Process licence at NZ$242.70 a month for a flow that legitimately needs more
9Not a fault: polling delay or a scheduleTrigger settings, and the time zone on a RecurrenceEvent triggers poll, typically every 1 to 5 minutes. A 9:00 schedule uses the time zone set in the trigger, which may not be yours

Source: Microsoft Learn, Fix connection failures in cloud flows, Troubleshoot Power Automate trigger problems and errors, Troubleshoot cloud flow errors and Understand platform limits and avoid throttling, all read on 30 September 2026. The Process licence price is Microsoft's NZ list.

The two causes that catch small firms most

The connection is a person. A flow that runs on someone's own sign-in stops when their password is reset, their licence is removed or their account is deleted. That failure pattern, and how to give a flow a connection that is not a person, is the whole subject of the flow that broke when the builder left.

Microsoft switched it off. A cloud flow that has failed continuously or been throttled for 14 days is turned off, and one that has not been triggered for 90 days can be too. Only the owners are emailed, so a firm can lose a flow without the admin ever hearing. The rules and the notice are set out below. Throttling, the other frequent cause, has its own numbers in Power Automate limits and throttling.

What happens when you turn it back on

This part surprises people. Microsoft's FAQ says that when a flow with a polling trigger, such as Recurrence, is turned on again, all the unprocessed events from the time it was off are processed. A webhook trigger only processes new events after it is switched on. So a flow that has been off for two weeks and watches a list may fire hundreds of times at once, sending emails or creating records. Before switching it on, check what it would do to the backlog, and delete the flow if you do not want the pending items handled.

Five minutes of prevention

  1. Add a second owner to every flow that matters, so the failure email reaches two people.
  2. Turn on failure alerts, or add a run-failed notification to a shared mailbox, so you hear about it on day one instead of day fourteen.
  3. Look at your connections once a quarter and put a diary date against any password that expires.
  4. Keep a list of what each flow does and who it affects. When one stops, that list tells you whether it is a Monday problem or a next-quarter problem.

Premium connectors, in flows or apps, need a paid licence: Power Automate Premium is NZ$24.30 per user per month on Microsoft's NZ list, and the cost worked out for a small firm is in Power Automate cost in NZ.

When a stopped flow is a sign of something bigger

  • It has stopped more than twice. A flow that keeps failing usually depends on one person's account or a connector that is wrong for the job. Fix the design, not the symptom.
  • Several flows stopped at once. That points to an environment or policy change, such as administration mode or a new data loss prevention rule, not each flow's own fault. If Outlook and Teams went down at the same time, check Microsoft's status first: what to do when Microsoft 365 itself is down.
  • Nobody knows what the flow was for. Then the inventory is the first job; your tenant has 200 flows and nobody owns them has the method.

Automatic suspension and the alerts you can miss

Microsoft's limits, checked on 30 September 2026, distinguish continuous failure, consistent throttling and consistently exceeding throughput limits. Each can turn a flow off after 14 days. Editing a flow resets the throughput-limit count because that limit applies to a version. Microsoft publishes no failure percentage. A flow never triggered for 90 days might also be turned off, with owners and co-owners warned 30 days beforehand; flows owned by premium-licensed users or covered by a capacity licence such as Process are exempt from that inactivity rule.

Per-run alerts only cover recognised, fixable causes. They go to owners and co-owners, never run-only users, environment admins or tenant admins, and one alert starts a 28-day cooldown for that flow. They are not enabled on every flow by default. A weekly digest covers failures regardless of cause. I would check the settings and add a run-after branch set to has failed that notifies a shared mailbox, rather than rely on the built-in emails alone.

If the portal does not explain a suspension, an admin can read its reason and time through the Power Automate Management connector or PowerShell. The reasons include AlwaysFailingDetected, AllActionsFailingDetected, ApiCallOverageDetected and NeverTriggeringDetected. Source: Microsoft Learn, Limits of automated, scheduled, and instant flows, Understand flow failure notifications and Troubleshoot a cloud flow, read on 30 September 2026.

Common questions

Why did my Power Automate flow stop working?

In the order I check: it was switched off or suspended, its trigger condition filters out every event, a connection expired, a licence lapsed, a data loss prevention policy or administration mode is blocking it, or it is being throttled. The flow's details page and 28-day run history show which within about ten minutes.

Why does Power Automate turn my flow off by itself?

Microsoft turns off a cloud flow that has failed continuously for 14 days, or been consistently throttled for 14 days, and can turn off one that has not triggered for 90 days. Only the flow's owners and co-owners are notified. Fix the cause, then turn it back on.

What happens when I turn a stopped flow back on?

A flow with a polling trigger, such as Recurrence, processes all the events it missed while it was off. A webhook trigger only handles new events after it is switched on. Check the backlog first, because a long-stopped flow can fire many times at once.

How do I fix a broken connection in Power Automate?

Go to More, then Connections, find the connection the flow uses and select the Fix connection link next to Status, then sign in again. Connections break when a password expires or when your organisation sets sign-in tokens to expire after a set time.

Who gets the email when a flow fails?

Only the flow's owner and co-owners. Personal flows email the creator, shared flows email all co-owners, and run-only users, environment admins and tenant admins are never sent per-run failure alerts. After one alert there is a 28-day cooldown for that flow.

Flow stopped and you cannot see why?

Free 30-minute call. Tell me what it did and roughly when it stopped, and I'll find the cause and tell you whether a fix or a redesign is the right size.

Book a free 30-minute call

Or ask me one question

Please enter your name.
Please enter a valid email.
0 / 500
Type your question so I know what to answer.

I use your email only to reply.

Find out on day one, not day fourteen.

I will set up failure alerts and a second owner on the flows that matter.

Christchurch-based. I reply within 1 working day.