Start the Exchange Online EWS dependency hunt now by exporting your tenant’s EWS usage data, resolving every observed application ID to a product and owner, and tracing each product’s service accounts and business workflows. Begin discovery before Microsoft’s phased retirement enforcement affects your tenant. This playbook focuses on dependencies that an Entra application inventory commonly misses: legacy Outlook integrations, native macOS apps, backup and archiving platforms, CRM and scheduling connectors, public-folder tools, and unattended scripts.

Exchange Online EWS dependency dashboard showing applications, risks, ownership, and Microsoft Graph migration steps.Start with observed EWS usage​

WindowsForum administrators discussing the EWS retirement consistently identify tenant usage reporting as the practical starting point. It shows observed activity, while configuration exports and staff interviews show only what teams already know or remember.
  1. Sign in to the Microsoft 365 admin center.
  2. Locate the tenant-specific reporting for Exchange Online EWS usage.
  3. Select a reporting period that covers a representative business cycle.
  4. Export or otherwise preserve the available usage data as a baseline evidence file.
  5. Record when the data was obtained and the period it covers.
  6. Create a working row for every observed application ID.
  7. Repeat the collection regularly through migration so that snapshots can be compared.
Do not assume that one quiet reporting period proves an application has been retired. Month-end processing, quarterly reporting, incident-response tools, restores, migrations, and executive scheduling workflows may generate EWS calls only occasionally.
EWS has supported mail, calendars, contacts, delegates, mailbox synchronization, notifications, public folders, and related workflow patterns. A dependency may therefore appear as a calendar-booking feature, archive collector, public-folder utility, or scheduled script rather than an application labeled “EWS.”
An Entra application export remains useful, but it is not a complete inventory. It may identify an application without explaining the product, mailbox scope, service identity, endpoint, or business process behind the observed activity. Microsoft 365 Message Center notices can help prompt investigation, but they should not replace a maintained dependency register.

Apply an evidence hierarchy before dispositioning a GUID​

No application ID should be approved, migrated, or dismissed from one source alone. Use this evidence hierarchy:
  1. Usage-report observation: Establish that EWS activity was observed in the tenant and preserve the relevant reporting evidence.
  2. Entra lookup: Resolve the application ID where possible and collect its display name, ownership, publisher, and sign-in context.
  3. Endpoint and service-account evidence: Find the host, scheduled task, automation platform, managed device, credential, or vendor service generating the activity.
  4. Owner confirmation: Have the technical and business owners confirm the workflow, impact, mailbox scope, and required remediation.
A GUID is ready for disposition only when these evidence layers agree. An Entra display name without operational evidence is insufficient, and an owner’s recollection without observed or technical evidence is not enough to close the item.

Turn application IDs into an accountable remediation list​

The useful deliverable is an owner-and-remediation register that can survive staff changes, vendor delays, and budget cycles.
Create one row for each distinct combination of application ID, product, workload, and operational owner:
FieldRequired information
Application IDGUID associated with observed EWS activity
Display nameEntra enterprise application or identified client
Product or scriptHuman-readable system name
EWS functionMail, calendar, contacts, delegates, synchronization, notifications, or public folders
Mailboxes accessedUser, shared, resource, archive, or public-folder scope
Authentication identityApplication, user, or service identity
Business ownerPerson accountable for the process
Technical ownerTeam supporting the integration
Vendor contactSupport case, account contact, or escalation route
Current statusInvestigating, exception candidate, testing, blocked, or retired
ReplacementMicrosoft Graph, vendor update, redesigned process, or retirement
Test evidenceDate, scope, and result of functional testing
Exception expiryDate temporary EWS access must end
For each application ID:
  1. Search Microsoft Entra admin center > Identity > Applications > Enterprise applications.
  2. Match the observed application ID to the corresponding application, if present.
  3. Record recognizable ownership, sign-in context, display-name, and publisher information.
  4. Determine which installed product, hosted service, script, or endpoint generates the calls.
  5. Identify the mailboxes and business functions involved.
  6. Contact the technical and business owners to confirm what fails if access stops.
  7. Obtain the vendor’s written migration position when a third-party product is involved.
  8. Assign a written disposition: migrate, update, replace, retire, or request a temporary exception.
  9. Give every exception an accountable owner and expiry date.
If an ID cannot be resolved in Entra, do not mark it harmless. Place it in an unresolved active usage queue and investigate endpoints, service identities, vendor portals, and workflow hosts.

Hunt the dependencies inventories miss​

Old Outlook-era integrations​

Review Outlook add-ins and companion applications that read mailboxes, create appointments, synchronize contacts, process delegates, or monitor folders. Ask about functions rather than protocols:
  • Does the tool file or classify messages?
  • Does it synchronize CRM contacts?
  • Does it create or modify appointments?
  • Does it use shared or delegated mailboxes?
  • Does it monitor a folder for incoming requests?
An add-in can have a modern interface while a server-side component still uses EWS. Test the actual workflow rather than accepting “Microsoft 365 supported” as proof of Graph readiness.

Native applications on macOS​

Native Apple Mail, Calendar, Contacts, Notes, and Reminders on macOS may need to be allow-listed until Apple updates them so that EWS is no longer required.
Inventory managed and unmanaged Macs, identify which native applications connect to Exchange Online, and assign the findings to an accountable endpoint or collaboration owner. Treat any allow-list entry as a temporary compatibility measure with a documented review date, not as a completed migration.

Backup and archiving products​

Backup, migration, archive, and eDiscovery-support products are high-risk because they often operate unattended and may access many mailboxes.
For each product, establish:
  1. Whether EWS is used for discovery, collection, restore, or several operations.
  2. Which mailboxes and mailbox types it accesses.
  3. Whether the deployed release supports Microsoft Graph.
  4. Whether migration requires a new connector, license, consent, or reauthorization.
  5. Whether historical backups remain restorable after changing connectors.
  6. How collection and restore failures appear in monitoring.
A successful backup dashboard is not sufficient evidence. Perform both a controlled collection and a controlled restore through the replacement path.

CRM, scheduling, and workflow connectors​

CRM synchronization, room-booking systems, interview schedulers, ticketing tools, and approval workflows may call EWS only when a particular feature runs. Low volume can therefore represent an important process.
Test representative operations:
  • Creating, updating, and canceling appointments
  • Synchronizing contacts
  • Reading or processing a shared mailbox
  • Handling delegate access
  • Detecting a new message or folder change
  • Recovering from an expired credential or denied permission
Include connectors configured inside SaaS administration portals. Their application identities and credentials may never have been documented by the Windows or Exchange team.

Public-folder tooling​

Give public folders their own workstream. Locate migration utilities, synchronization products, archive collectors, line-of-business readers, and scripts that enumerate or copy public-folder content.
Require owners to demonstrate the workflow under test conditions. Access to ordinary mailbox folders does not prove that a replacement supports the required public-folder operation.

Scripts and forgotten service identities​

Search scheduled tasks, automation servers, runbooks, integration hosts, deployment records, and source repositories for Exchange endpoints, EWS libraries, SOAP requests, stored credentials, and mailbox-processing code.
Review service identities for:
  • Noninteractive accounts with Exchange-related names
  • Accounts exempted from normal sign-in controls
  • Shared credentials held by multiple teams
  • Tasks running under former employees or generic administrators
  • Scripts accessing shared mailboxes or calendars
  • Jobs running only weekly, monthly, quarterly, or during incidents
A dormant script remains a dependency if the business expects it to work. Test it or obtain written approval to retire it.

Decide what to migrate, allow, or remove​

WorkloadPreferred actionRequired proof
Supported clientUpdate and retestEWS activity ends without feature loss
Outlook-era vendor integrationInstall a Graph-capable releaseEnd-to-end workflow succeeds
Native macOS applicationsTrack the required update; use a temporary exception if necessaryManaged-client test and documented exit date
Backup or archive platformMove to a supported connectorCollection and restore tests succeed
CRM or schedulerReauthorize, upgrade, or replaceCalendar, contact, and delegate tests pass
Public-folder toolValidate explicit replacement supportRequired public-folder operation succeeds
Custom scriptRewrite, replace, or retireGraph test or signed retirement approval
Unknown GUIDInvestigate; do not approve blindlyIdentified product, owner, identity, and scope
EWSAllowedAppIDs identifies applications permitted to continue through their Microsoft Entra application IDs. Treat an allow-list entry as an exception control, not as evidence that the application is ready for EWS retirement.
Before approving an ID, confirm that it belongs to the expected product, determine what it accesses, document the business reason, and set an exit date. Do not turn an unresolved GUID into an indefinite exception merely because a workflow breaks.

Test without losing control of the evidence​

Microsoft may conduct temporary EWS “scream tests” that expose hidden dependencies. A failure during a test can reveal systems that did not appear in interviews or inventories.
Tenants that set EWSEnabled=True are excluded from those tests. Understand that consequence before changing the setting: forcing EWS on may preserve service temporarily, but it can also remove a useful discovery signal.
For every migration:
  1. Preserve the current usage evidence for the application.
  2. Define the exact user and system operations to test.
  3. Move a controlled pilot to the replacement.
  4. Test normal operations, delegated access, failures, and recovery.
  5. Check product and Microsoft 365 monitoring for errors.
  6. Collect a later EWS usage snapshot after enough time has passed for the tested workflow to run.
  7. Confirm that old activity has stopped or is declining as expected.
  8. Obtain business-owner acceptance.
  9. Remove the exception only after verification.
  10. Retain targeted rollback instructions through post-cutover monitoring.
Rollback should restore only the affected application. Do not respond to one failed integration by broadly enabling EWS for every client.

Troubleshoot unresolved or persistent activity​

If EWS activity continues after migration:
  1. Confirm that every production instance received the new connector or configuration.
  2. Check disaster-recovery hosts, test systems, old virtual machines, and secondary automation nodes.
  3. Look for cached credentials or scheduled tasks still using the former path.
  4. Verify that the vendor’s desktop, server, and hosted components were all addressed.
  5. Recheck shared service identities for use by another workflow.
  6. Ask the owner to run the business process while endpoint and identity evidence is collected.
  7. Keep the item open until new usage observations, technical evidence, and owner confirmation agree.
If no owner can be found, escalate the application as an operational risk. Lack of ownership is not evidence that a dependency is safe to block.

Frequently Asked Questions​

Is an Entra app-registration export enough?​

No. It may identify registered applications without explaining the product, service identity, mailbox scope, endpoint, or business workflow behind observed EWS activity. Combine usage observations, Entra lookup, endpoint or service-account evidence, and owner confirmation.

Does low EWS call volume mean an application is unimportant?​

No. Scheduling, restore, public-folder, reporting, and incident-response workflows may run only occasionally. Assess business impact and function, not call count alone.

Can EWSAllowedAppIDs be used as the final solution?​

No. It is a temporary control for explicitly permitted applications while they are updated, migrated, replaced, or retired. Every allow-listed ID needs an owner, business justification, review date, and exit date.

Does this retirement apply to on-premises Exchange?​

The Microsoft Graph migration guidance discussed here targets Exchange Online and hybrid deployments accessing Exchange Online. Microsoft Graph does not provide Exchange on-premises support as a replacement for on-premises EWS.
Before phased retirement enforcement affects the tenant, every observed EWS dependency should have a confirmed product, functional purpose, technical owner, business owner, tested remediation path, and time-limited exception where migration cannot yet be completed.

References​

  1. Primary source: learn.microsoft.com
  2. Independent coverage: techcommunity.microsoft.com
  3. Primary source: WindowsForum