Most Exchange Online problems fall into five buckets: Microsoft-side service incidents, tenant configuration or policy blocks, client or authentication failures, protocol-specific issues, or mail-flow and connector faults. Split first by scope, then by protocol, then by recent changes. Misclassifying the bucket wastes outage time, delays mail, and can break access for every mailbox or only one access path. This guide ties symptoms to checks, error patterns, and fixes for Outlook, OWA, ActiveSync, SMTP, connectors, DNS, and transport.
Start with tenant-specific Service health, then narrow the blast radius: one user, one app, one protocol, one site, or the whole tenant. That order tells you whether to wait on Microsoft, correct tenant settings, collect proof, or open a support case with evidence that support can use.
What fails in Exchange Online, exactly?
Exchange Online can fail at the service, identity, DNS, client, transport, filtering, or protocol layer, and users often see the same symptom from very different causes. A mailbox that will not open, a phone that will not sync, and a missing message can each point to a different layer, so define scope and protocol before you try to fix anything.
User-facing symptoms by layer
Mailbox access failures usually show up first in Outlook desktop, Outlook on the web, or mobile mail. Calendar issues may appear as stale free/busy data, missing meetings, or delayed updates even while mail delivery still works.
Mail-flow problems look different. Users may send successfully but never receive, receive internal mail but not external mail, or find messages in quarantine because Microsoft Defender for Office 365 policy caught them.
Protocol-specific failures are narrower. A line-of-business app using Microsoft Graph may fail while Outlook works. EWS-based add-ins may stop working after auth or consent changes. SMTP AUTH can fail for printers, scanners, or older applications while mailbox access stays normal.
Fast scope check
Start by placing the problem in one box: one user, many users, one location, one client type, or tenant-wide. Then ask whether the issue affects mailbox access, mail flow, calendar, or one protocol such as ActiveSync, EWS, Graph, SMTP AUTH, or Exchange Online PowerShell.
- One user on one device points first to client state, token state, local profile, or device policy.
- Many users on one network points to identity, Conditional Access, proxy, firewall, or DNS.
- All users in one protocol points to a policy or protocol setting, not a full Exchange outage.
- Tenant-wide symptoms across Outlook, OWA, and PowerShell point first to service health or tenant identity controls.
Why similar symptoms split into different causes
“Outlook disconnected” can mean a stale profile, failed Modern authentication, broken Autodiscover, or a Microsoft incident affecting client connectivity. “Mail not received” can mean transport delay, rejected mail, bad MX records, quarantine action, or a user checking the wrong client folder.
Suppose Outlook on the web opens, but Outlook desktop shows repeated password prompts for thirty users after a Conditional Access change. That pattern points to auth policy or token-path trouble, not a transport issue and not a broad Exchange service failure.
How do you archive old emails in Exchange Online?
Archive old mail by enabling the archive mailbox and then assigning a retention policy that moves messages after a set age, such as 2 or 5 years. The supporting guide walks through the setup steps, how to make the archive visible to users, and how to verify that mail is actually being moved instead of staying in the primary mailbox.
This matters because archiving helps keep user mailboxes smaller, reduces the chance of hitting storage limits, and supports retention requirements. It also covers common mistakes, like skipping policy assignment or not checking whether users can access archived messages, which can leave the archive technically enabled but functionally useless. See the archive mailbox setup guide for the full process.
How do you tell if it is Microsoft or your tenant?
The fastest split is the Microsoft 365 Service health dashboard plus tenant evidence from sign-ins, message trace, and protocol scope. If Service health shows an Exchange Online incident that matches the symptom and timing, wait or use a workaround. If it is clear, treat tenant, policy, DNS, or client causes as more likely until the evidence says otherwise.
Read Service health correctly
The Microsoft 365 Service health dashboard is often the first check because it is tenant-specific and separates incidents from advisories. An incident means a service-side problem with active impact. An advisory often means degraded behavior, narrow scope, or a known issue with a workaround, and it may not affect every tenant.
Read the affected service carefully. Exchange Online incidents are tracked separately from other Microsoft 365 services. A Teams or Entra issue may still explain Exchange sign-in failures, but it is not the same as an Exchange transport or mailbox-access incident.
Use tenant signals when Service health is clear
If Service health is clear, check whether the pattern is tenant-wide or limited to one path. OWA failing for every user suggests identity or service issues. Outlook desktop failing while OWA works can suggest client auth, Autodiscover, profile, or cached-state issues. External senders failing while internal mail works points to DNS, connectors, filtering, or accepted-domain configuration.
Message trace separates mail flow from client complaints. Sign-in logs separate identity blocks from mailbox or transport failures. Exchange Online PowerShell connection status helps confirm whether admin access paths are also affected.
Wait, workaround, or fix in the first 10 minutes
- Check Service health first. Open your tenant’s Microsoft 365 Service health dashboard, verify that the affected workload is Exchange Online, and compare the posted start time with the first user reports. If the incident or advisory matches both symptom and scope, stop changing tenant settings and follow the posted updates.
- Define the blast radius. Test one affected user and one unaffected user in Outlook desktop, OWA, and the mobile mail client. That comparison shows fast whether the break follows the user, the client, the network path, or the whole tenant.
- Pick the likely layer. Use one quick proof source that matches the symptom: Entra sign-in logs for authentication failures, Exchange message trace for mail-flow complaints, DNS checks for MX or Autodiscover issues, and Exchange Online PowerShell connection attempts for admin-channel failures. That first split prevents random fixes.
- Decide the next action. Wait on Microsoft for a matching incident, correct tenant settings for clear local misconfiguration, or collect evidence and escalate when the issue is tenant-wide with no obvious local cause. Avoid profile rebuilds until you know the failure layer.
Exchange problem matrix
| Symptom | Likely layer | Fastest verification step | Next fix | Escalation trigger |
|---|---|---|---|---|
| OWA and Outlook fail for many users at once | Service or identity | Check tenant Service health and Entra sign-in logs | Wait for matching incident or correct tenant-wide auth policy | Open Microsoft case if tenant-wide, no posted incident, and logs show broad failures for 30+ minutes |
| Outlook prompts for password repeatedly, OWA works | Client or auth | Test Outlook with modern auth sign-in and review Conditional Access result | Clear stale credentials, reauthenticate, review MFA and CA policy | Escalate if the pattern hits many users after no policy change is visible |
| New Outlook profile cannot auto-configure mailbox | Autodiscover or DNS | Validate Autodiscover DNS and test profile creation on another network | Correct Autodiscover records or endpoint discovery path | Escalate if correct DNS still fails across multiple networks and users |
| Only one user cannot open mailbox in Outlook or OWA | Identity or mailbox state | Review sign-in logs and mailbox/service assignment | Fix account state, license, or mailbox provisioning issue | Escalate if license and sign-ins are normal but mailbox remains inaccessible |
| Internal mail works, external inbound mail missing | DNS or transport | Check MX records and run message trace for expected messages | Correct MX routing, accepted domains, or inbound connector settings | Escalate if DNS is correct and no inbound handoff reaches Exchange |
| Messages delayed or stuck in transport | Service or transport | Use message trace to confirm delayed status and compare timing with Service health | Wait on incident or review connectors and transport rules | Escalate when trace shows prolonged delay without matching advisory |
| Messages delivered to quarantine unexpectedly | Filtering | Review quarantine and Defender policy action | Release, adjust policy, or create approved exceptions where justified | Escalate if false positives are broad and policy configuration looks unchanged |
| Phones stop syncing, Outlook and OWA still work | ActiveSync, device policy, or auth | Check device partnership status and sign-in events for mobile clients | Refresh partnership, review device access rules, and reauthenticate | Escalate if multiple device types fail with healthy sign-ins and no policy change |
| Graph or EWS app fails while mailbox access is normal | Protocol or consent | Check app permissions, token acquisition, and API-specific errors | Renew consent, fix auth flow, or update deprecated dependency | Escalate if app tokens are valid and failures are tenant-wide for the same endpoint |
| Printer or app cannot send mail | SMTP AUTH or connector | Confirm whether SMTP AUTH is enabled for the mailbox or tenant and test connector path | Enable the required send path or move to authenticated submission/connector design | Escalate if settings are correct and submissions fail from multiple sources |
| Exchange Online PowerShell connection fails for admins | Module, auth, or network | Verify Exchange Online module version, sign-in prompt, and outbound network path | Update module, satisfy modern auth, and allow required endpoints | Escalate if multiple admins fail on updated modules and healthy auth paths |
| Calendar updates slow or missing, email works | Client, service, or add-in/protocol | Compare Outlook desktop with OWA calendar behavior and check Service health | Fix client cache/add-in issue or wait on service incident | Escalate if OWA and Outlook both show tenant-wide calendar inconsistency |
| Only one office loses Outlook connectivity | Network or DNS | Test OWA, DNS resolution, and proxy/firewall from that site | Correct local egress, proxy auth, SSL inspection, or resolver issues | Escalate if the same users fail from other networks too |
Why can’t Outlook connect to Exchange Online?
Outlook connectivity failures usually come from four places: Modern authentication, Autodiscover, client profile state, or cached local data. If OWA works while Outlook does not, focus first on the desktop client path, because the mailbox and service may be healthy even while Outlook keeps prompting or stays disconnected.
Outlook desktop failures
Repeated password prompts usually point to token or policy trouble. A disconnected status with no prompt can point to network inspection, endpoint resolution, or a damaged profile. Errors during first-time setup point more strongly to Autodiscover or DNS.
Check whether affected users can sign in to OWA in the same browser session. If yes, mailbox availability is likely fine and the Outlook path deserves attention first.
OWA path failures
OWA issues help because they remove local Outlook profile variables. If OWA also fails, review the browser sign-in path, Conditional Access result, MFA challenge, and any broad incident affecting Exchange Online or identity services.
Picture a branch office where Outlook desktop fails for 80 users, but OWA opens from mobile data on the same accounts. That split points away from mailbox corruption and toward local network controls, DNS, or SSL inspection on that office path.
Autodiscover, cached data, and client version mismatch
Autodiscover is still central for Outlook profile creation and endpoint resolution. Incorrect DNS or a redirected discovery path can stop profile setup even when the mailbox itself is healthy. Older Outlook builds can also fail modern auth or service endpoint expectations after tenant changes.
- Test OWA access. Sign in with the same affected account through Outlook on the web. A successful OWA session means the mailbox is present and narrows the issue to client auth, Autodiscover, or local Outlook state.
- Review sign-in outcome. Check whether the account receives a successful token and whether Conditional Access or MFA blocks the client. This shows whether Outlook is failing before mailbox access starts.
- Validate Autodiscover path. Confirm the Autodiscover DNS record and create a new profile only after the discovery path looks correct. Rebuilding a profile before this check wastes time and often reproduces the same failure.
- Check client state. Update Office, clear stale credentials, and test without old cached mode data if the failure is limited to Outlook desktop. Use profile recreation only after auth and discovery checks point there.
How do you fix sign-in and Modern authentication failures?
Modern authentication failures usually come from Conditional Access, MFA, expired or stale tokens, disabled legacy paths, or account-state problems. The key is to identify whether the block is policy-driven, client-driven, or protocol-driven, because Outlook, mobile clients, PowerShell, EWS, and Graph do not all fail the same way.
Conditional Access, MFA, and policy blocks
Check the sign-in result before changing mail settings. Conditional Access may require a compliant device, approved app, known location, or MFA claim that the current client cannot satisfy. This is common when Outlook mobile works but a desktop build or older app does not.
If a recent policy change lines up with the outage start, reverse it or exclude one test account long enough to prove the cause. That is faster than rebuilding multiple clients.
Legacy auth dependencies and account state
Some older tools still depend on legacy protocols or older auth behavior. When legacy auth is blocked, Outlook may still work while SMTP AUTH, older EWS workflows, or device mail apps fail. Also check for expired passwords, disabled accounts, missing licenses, or mailbox provisioning delays.
Exchange Online PowerShell sign-in failures
Exchange Online PowerShell depends on the module version, modern auth flow, and network reachability. A broken module or blocked sign-in prompt can look like an Exchange service problem even when mailbox access is fine.
- Read sign-in logs. Capture the exact error code, Conditional Access result, and failure reason for one affected user or admin. This shows whether the denial came from identity policy, MFA, token, or client incompatibility.
- Check account state. Confirm the user or admin account is enabled, licensed where required, and not blocked by expired password or risky-sign-in controls. Mailbox problems rarely cause an identity denial on their own.
- Verify protocol dependence. Identify whether the failing path is Outlook, ActiveSync, EWS, Graph, SMTP AUTH, or PowerShell. A protocol-specific failure often points to a setting, permission, or app-consent issue rather than a broad sign-in outage.
- Update the client or module. For Exchange Online PowerShell, use a current module build and complete the modern auth prompt without local browser restrictions. For Outlook and mobile apps, update the client before deeper protocol work.
What should you check first when mail stops flowing?
Check message trace first, then quarantine and transport settings. Message trace shows whether Exchange Online accepted, delayed, rejected, redirected, or never saw the message, which is the fastest way to separate transport failure from a user looking in the wrong place or a client sync delay.
Message trace and transport status
Use message trace for the sender, recipient, and a time window in UTC that brackets the complaint. Compare internal and external tests. If Exchange never received the message, the problem may be upstream DNS, sender-side routing, or connector configuration rather than Exchange transport.
If trace shows delivered, the issue is usually client view, rules, quarantine release path, or mailbox processing. If trace shows failed or pending, focus on transport.
Quarantine, Defender, and rules
Review Microsoft Defender for Office 365 quarantine and policy actions for false positives or policy-driven blocks. Then check transport rules, connectors, and accepted domains. A connector mismatch or bad accepted-domain setting can break only one direction of mail while users still send internally.
Say your company changed its MX record six hours ago and new external mail stopped, but internal mail and OWA still work. Message trace showing no inbound messages after the cutover points first to public DNS or mail routing, not a mailbox database issue.
External complaints versus internal failures
Inbound external failures and outbound external failures are different cases. Inbound trouble points first to MX, connectors, and filtering. Outbound rejection points first to SMTP submission path, relay design, connector scope, or sender-authentication alignment.
- Run message trace. Search by sender, recipient, and a UTC time window that includes the reported send time. Note whether the message is delivered, expanded, redirected, quarantined, pending, failed, or absent from trace.
- Check quarantine next. Review Defender quarantine for the recipient and policy action taken. This separates security filtering from transport rejection and avoids changing connectors for a filtering-only event.
- Inspect transport settings. Validate connectors, accepted domains, and any recent transport rule changes that match the symptom direction. Inbound and outbound failures often have different root causes even when users describe both as “mail down.”
- Compare internal and external tests. Send one internal and one external message with known timestamps. Different outcomes narrow the fault to routing, filtering, or public DNS rather than mailbox access.
Why are mobile devices or apps failing while mailboxes work?
When mailbox access works in OWA or Outlook but phones or apps fail, the issue usually sits in ActiveSync, app permissions, token handling, SMTP AUTH, or device policy. This is rarely a full Exchange outage, because the mailbox and core service path are already proven healthy.
ActiveSync sync failures
Exchange Online ActiveSync failures often show up as repeated password prompts, mail that stops updating, or only calendar and contact sync breaking. Check whether the device partnership is healthy, whether the device is blocked by access rules, and whether the account can still complete modern auth on that mobile client.
Graph, EWS, and mailbox app access
App-specific mailbox failures often trace to missing consent, changed permissions, expired secrets, or deprecated dependencies. If Graph or EWS calls fail but the same user can access Outlook, do not spend time on transport or MX. Focus on app auth and the exact API path.
SMTP AUTH and legacy send paths
Many older applications send by SMTP AUTH and fail after security tightening. The mailbox may receive mail normally while the app cannot submit. Confirm whether SMTP AUTH is enabled where needed and whether a connector-based relay design is the better fit.
Which DNS and domain settings mimic an Exchange outage?
Bad DNS can look like a Microsoft outage by breaking inbound mail, Autodiscover, or sender trust while Exchange itself stays healthy. Check MX, Autodiscover, SPF, and any related routing records before assuming the service is down, especially after domain, provider, or connector changes.
MX and mail-routing errors
Incorrect MX records, stale TTL propagation, or routing mail to a retired gateway can stop external delivery while internal mail works. When message trace shows no inbound arrival, public DNS is a top suspect.
Autodiscover DNS and client discovery
Autodiscover DNS issues break profile creation and endpoint resolution. Outlook may fail during setup, or a rebuilt profile may never finish. This is one of the most common ways DNS problems masquerade as a broad Exchange connection failure.
SPF and related sender-authentication effects
SPF does not control mailbox access, but it can affect deliverability, especially outbound complaints and spam placement. If users say “mail is broken” after a mail-routing change, check whether sender-authentication records still match the new path.
Separate public DNS faults from Microsoft faults by testing from multiple resolvers and networks. If only one office or one ISP path sees bad results, local caching or resolver issues may be involved.
When should you escalate to Microsoft support?
Escalate when the issue is tenant-wide without a matching posted incident, when protocol failures persist after you verify the likely tenant settings, or when mailbox access, transport, or admin channels fail with evidence that rules out local policy, DNS, and client causes. Good evidence shortens case routing time.
Evidence to collect before opening the case
Collect affected users, exact error strings, timestamps in UTC, screenshots, correlation or request IDs, message trace output, sign-in log results, and whether OWA, Outlook, mobile, and PowerShell each work or fail. Include recent tenant changes: Conditional Access, connectors, DNS, transport rules, app consent, and mobile policy.
Escalation triggers by symptom type
- Mailbox access: many users affected across OWA and Outlook, no matching Service health incident, sign-ins look normal or inconclusive.
- Mail flow: message trace shows repeated unexplained delay or rejection after connectors, domains, and quarantine have been checked.
- PowerShell: multiple admins on current modules cannot connect from different networks with healthy identity prompts.
- Protocol/app access: Graph, EWS, ActiveSync, or SMTP AUTH fails across multiple users or apps after local settings and permissions are verified.
What to include so support can route faster
- Summarize the scope. State whether the problem affects one user, one site, one protocol, or the whole tenant. Include the first known UTC timestamp and whether the problem is ongoing or intermittent.
- Attach exact errors. Paste the full visible error text, request or correlation IDs, and a screenshot from the failing client or admin tool. Short paraphrases lose the clues support needs for routing.
- Provide proof artifacts. Add message trace results, sign-in log outcomes, and the Service health status seen at the time of impact. These three artifacts usually answer whether the fault is transport, identity, or service-side.
- List recent changes. Include any edits to Conditional Access, DNS, connectors, transport rules, Defender policy, SMTP AUTH settings, app permissions, or module updates. A support engineer will ask for this anyway, so sending it first saves a round trip.
Frequently asked questions
Why is Exchange Online not working today?
Start with the tenant’s Microsoft 365 Service health dashboard and compare the incident or advisory timing with your first failures. If Service health is clear, split by scope and protocol: OWA, Outlook, mobile, PowerShell, and message trace results usually show whether the cause is service-side, tenant policy, DNS, or client-specific.
How do I tell if the problem is Microsoft’s service or my tenant?
A matching Exchange Online incident or advisory in Service health is the strongest sign that you should wait or use a workaround. If no matching event exists, compare OWA, Outlook, mobile, and PowerShell, then check sign-in logs and message trace. Mixed results usually point to tenant policy, DNS, client, or protocol settings.
What should I check first when mail stops flowing in Exchange Online?
Run message trace first for the reported sender, recipient, and UTC time window. It shows whether Exchange accepted, delayed, rejected, quarantined, or never saw the message. Then review Defender quarantine, connectors, accepted domains, and MX records to separate filtering, transport, and public-routing causes.
What logs or tools help diagnose Exchange Online problems?
The most useful set is Service health, Entra sign-in logs, Exchange message trace, Defender quarantine, Exchange Online PowerShell, and public DNS checks for MX and Autodiscover. Capture exact error strings, correlation IDs, screenshots, and UTC timestamps so the evidence lines up across tools and support channels.
When should I escalate an Exchange Online issue to Microsoft support?
Escalate when the issue is tenant-wide without a matching incident, when multiple protocols fail after tenant settings are checked, or when message trace and sign-in logs show unexplained failures that persist. Include affected users, scope, exact errors, UTC timestamps, traces, screenshots, and recent configuration changes.