For data owners and authorised Microsoft 365 administrators
Use the actual workload, policy and configured recovery service. A subscription, feature label or completed job does not establish every item, version or access arrangement.
Step 1 of 9
In this guide
Identify the missing data and preserve current work
Identify the missing data and preserve current work
Record the actual work account/tenant, data owner, workload and source location. Is this a SharePoint library, OneDrive account, mailbox, or another Microsoft 365 service? Note the file/item, expected version, last known good time and what changed. A Teams file may live in a SharePoint or OneDrive location; do not assume all chat, meetings and files share one recovery method.
Preserve open and unsynchronised work through the approved process. Check the genuine source with the owner before deciding it was deleted: a missing shortcut, changed permission, different account or incomplete sync can look like data loss. Use the file source and sync guide when location/access is uncertain.
Act promptly when data is missing or damaged because recovery options can expire. For suspected compromise or widespread unwanted changes, involve the security and recovery owners before normal bulk changes. Do not empty bins, remove users/sites, delete backups, repeatedly synchronise damaged copies or restore a whole library merely to see what happens.
Identify the real workload/source, expected content and last good time. Preserve current work, distinguish access/sync from deletion and escalate time-sensitive loss or compromise.
Next: Distinguish service resilience, retention and configured backupDistinguish service resilience, retention and configured backup
Distinguish service resilience, retention and configured backup
Do not begin with the assumption that Microsoft 365 provides no recovery, or that a subscription automatically gives unlimited backup of everything. Service resilience, recycle bins, version history, retention and configured backup address different needs. The actual product, settings, covered workloads and available historical copies determine what can be recovered.
Sync keeps copies aligned and can carry deletions or unwanted edits to other locations. Versions and bins can help within their actual scope and lifetime. Retention may preserve information under a policy or legal/business requirement, but it is not the same as a user-controlled operational restore plan. Do not change retention or hold settings to troubleshoot missing files.
A backup arrangement needs defined protection scope, history, access, isolation from relevant failure/deletion risks and a usable restore method. It may use a Microsoft or another supported service; a brand name, licence or paid invoice does not establish coverage. Identify the actual configured arrangement rather than infer capabilities or provider architecture.
Match resilience, sync, versions, retention and backup to their actual purposes and limits. Neither no-recovery nor everything-backed-up is a safe blanket assumption.
Next: Inventory the recovery options already availableInventory the recovery options already available
Inventory the recovery options already available
Ask the authorised owner which recovery mechanisms already covered this source before the incident: item/version history, site or account recovery, retained content and configured backup. Record the relevant time range, content types and permissions. Check whether the expected item and version can actually be found, rather than only whether a feature is enabled.
For SharePoint/OneDrive, distinguish a missing item from a deleted library, entire site/account, or a deleted user’s lifecycle. Different scopes and related group resources can have different rules. The current item-recovery guidance and responsible administrator can identify the applicable route.
For configured backup, confirm actual inclusion and completed usable history before the required time. Not set up, processing, paused, excluded or newly added sources may not have the needed recovery point. Starting backup now cannot recreate a historical version that was never captured or retained elsewhere. Avoid fixed universal day counts when the workload, policy or state has not been established.
Locate actual available historical content and the applicable workload/role. A feature setting or newly started job cannot establish an older recoverable copy.
More checks6 detailed checks
Assess permanently deleted content with the owner
Assess permanently deleted content with the owner
Confirm the exact source and what “permanently deleted” means in this case. If you have permission, inspect the appropriate site/account recycle bin and selected item details without emptying it. If the item is unavailable, ask the site/data administrator promptly about the applicable later-stage, site/account, retention or existing backup options. A desktop bin and a SharePoint site bin are not interchangeable.
Use the owner’s supported process for the actual workload and time. An item no longer visible to an end user may have another permitted recovery route, but there is no promise it does. Do not claim an extra universal support-backup window, indefinite retention or that the entire tenant can always be rolled back. A file undelete utility cannot reconstruct a cloud copy that the service no longer holds.
If no applicable retained copy or backup contains the required data, be honest about the limit and document the assessment. Do not recreate an empty file and describe it as recovery, reset the account, remove holds or purchase a service on an unconditional recovery promise. If a suitable copy is found, plan the narrow restore and its effect on current work before confirming it.
Have the owner assess the actual available item/admin/retention/backup paths promptly. No visible copy is not an automatic final answer, but recovery requires a real available source.
Next: Plan a scoped restore without overwriting current workCheck backup scope, history, access and gaps
Check backup scope, history, access and gaps
Inventory the required mailboxes, sites, OneDrive accounts and other work, then compare them with the real backup scope and exclusions. Record newly created sites/users, shared/special accounts and offboarded sources; do not assume selection of one workload includes every Microsoft 365 app, message, setting or permission. Confirm product/version/entitlement limits with the responsible provider.
Inspect recent usable restore points, historical range, unsuccessful or stale jobs and the response owner. Microsoft’s backup policy documentation illustrates why selected scope and active state matter; it is not a statement about your particular service. Verify actual behaviour for automatic inclusion, exclusions and stopped protection.
Confirm who can access and delete backups, how that access is protected, how recovery works if the normal account is unavailable, and the applicable storage/retention requirements. Avoid unsupported claims about independent failure domains or deletion resistance. An independent recovery design needs evidence, including an appropriate restore exercise, not a second sync folder or a green job alone.
Compare actual protected sources, usable history, exclusions and privileged access with requirements. Resolve stale/new/offboarded gaps and verify recovery rather than rely on a dashboard or invoice.
Next: Plan a scoped restore without overwriting current workPlan a scoped restore without overwriting current work
Plan a scoped restore without overwriting current work
If a real recovery is needed, agree the authorised owner, exact source/items, required good time and destination before confirmation. Preserve current content and useful changes through the approved process. Consider a separate approved location or selected-item recovery where the actual service supports it; keep access appropriate and do not assume every product restores the same fields or destinations.
Microsoft’s restore documentation describes original-location replacement and other available scopes. A library rollback can undo wider activity than a single-file recovery. Read the actual preview/impact and reconcile late edits; do not use broad rollback for one missing item without assessing the effect.
For suspected ransomware or ongoing unauthorised access, coordinate containment and a healthy recovery point with the security owner. Restoring data does not remove the cause of an incident. Do not overwrite production, expose restored private content or delete old recovery points as a routine test. If the available source does not meet the need, record that limit before proceeding.
Agree the selected source/time/destination and preserve current work. Understand replacement and wider rollback effects; recovery must not create a new loss or hide an ongoing incident.
Next: Verify restored content and the future recovery planVerify restored content and the future recovery plan
Verify restored content and the future recovery plan
For a real authorised recovery, inspect representative restored items against the required date/version and content. Confirm files open in the intended app, expected structures and necessary metadata are present, and permissions are appropriate. For mail or other workloads, check the actual supported item types and result rather than infer completeness from restored files.
For a planned backup test, use a harmless approved sample and isolated destination supported by the provider. Do not delete important originals, overwrite a live site or submit private data as a demonstration. Record the recovery point, scope, duration, result and limitations; one sample does not prove every workload or future incident.
Reconcile current work and unresolved items with the data owner, and confirm future inclusion, usable history, access and failure-response owners. Keep retention and old-source disposal on their authorised schedule. If the needed content is absent, say so and escalate the actual gap; do not claim all data protected because the restore button or dashboard is available.

Confirm usable intended content, version and access, with representative scope and limits recorded. Keep backup coverage and response ownership maintained; one restore is not universal recovery proof.
Back to the symptom checkerGet help with the exact data and recovery scope
Get help with the exact data and recovery scope
Use the authorised Microsoft 365/data administrator and actual backup provider to assess the workload, historical copies, privileges and safe restore. Supply the source/site/account reference through an appropriate private channel, approximate loss/last-good time, item type and minimal relevant status/error. Do not send passwords, recovery codes, unrestricted exports or customer files in an initial enquiry.
Our Microsoft 365 backup service can start a scoped discussion about coverage and recovery requirements. Confirm the actual included data, history, access, retention, testing and response responsibilities before work. Do not assume an unconfigured backup exists or accept a promise to recover data without an available source.
For active compromise, keep incident response involved. For source/sync confusion, resolve the actual owner and copies first. Record what can be recovered, what remains unknown or unavailable and the next responsible action rather than call a whole tenant protected after one successful job.
Agree the exact workload and available recovery source with authorised owners. State gaps and limits plainly, preserve privacy and separate incident response from data restoration.




