With a corrupted Word document, the original file is evidence. Make a duplicate first, test repair on the duplicate, and keep notes on which copy came from which folder or backup.
First move: For a corrupted Word document, make an untouched archive copy, then run repair only on a duplicate.
The five-minute safety pass
Make the machine boring before recovery starts: no cleanup, no bulk restore, no replacement save. Record the last known device, account, folder, and time so the next check has a direction.
The record turns panic into triage. Match the event to the evidence source before opening random results or restoring a folder.
Follow the file from app to cloud to backup
Check the most likely owner of the newest copy first, then widen the search. For this case, start with:
File > Info > Manage Document > Recover Unsaved Documents%AppData%\Microsoft\Word\and theUnsavedFilesfolder for.asdor.wbkfilesFile > Info > Version Historywhen the file was saved to OneDrive or SharePoint- Recycle Bin, then File History or Time Machine if a saved copy was deleted
When a candidate looks possible, copy it before testing. Add a source label like autorecover, trash, version-history, or file-history so you can retrace the path later.
Move candidates into a safe folder
- Sort candidates by modified time, then by file size.
- Copy each likely file into the holding folder.
- Open copies read-only when the app allows it.
- Compare the first page, the last edited section, and any formulas, comments, or images that matter.
- Restore only after the best copy is clear.
This extra copy step is what keeps a reversible mistake reversible. If one candidate opens badly, you still have the untouched source. If two versions both contain useful work, you can compare instead of choosing blind.
The practical route for this file type
With Office, do not let the local sync folder make the decision alone. The desktop app, AutoRecover location, and browser version history can each hold a different clue.
If the file matters more than the device, spend the first five minutes preserving options rather than hunting for one perfect button.
A realistic recovery sequence
The file opens with an error after a crash. Make two copies: one untouched archive and one working copy. Try the built-in repair on the working copy, then compare file size and visible content. If the first repair produces blank pages or broken formulas, stop using that result and return to the untouched copy.
Which path to try first
| What you see | Best first check | Risk to avoid |
|---|---|---|
| The app crashed before a save | AutoRecover, unsaved files, and temp locations | Reopening and saving a blank replacement |
| The file was deleted from a synced folder | Web Trash or recycle bin for the cloud service | Letting the synced deletion become the only story |
| The content is older than expected | Version history and downloaded copies | Restoring over the live file before comparing |
| The file opens with errors | Duplicate first, repair only the duplicate | Damaging the only original with repeated repair attempts |
| Search finds many odd names | Sort by date, extension, and size | Opening every result and accidentally saving changes |
Mistakes that erase useful clues
- Restoring the first visible result because the filename looks right.
- Trusting Recent files instead of checking the actual storage location.
- Opening candidates directly from a temp folder and then saving over them.
- Running third-party recovery tools before copying cloud and backup evidence.
- Waiting until cloud Trash, version history, or backup retention windows expire.
The point where guessing gets expensive
Stop if the missing document is legally important, business-critical, the only copy of a major project, or stored on a drive that is making noise or disconnecting. Also stop if every repair attempt makes the file smaller or less readable. At that point, a clean handoff with copied candidates and notes is more valuable than another random tool.
Keep a short version history of your own
Before closing tabs, write down the recovery trail. Include the expected file path, the account used, the candidate names, and whether each came from recovery folders, Trash, version history, backup, or local search.
This record helps even when the recovery works. If the restored document later turns out to be incomplete, you can return to the candidate list instead of starting over. It also makes a professional handoff cleaner because the next person can see what was already checked and which version looked strongest.
Sources and further reading
- Microsoft Support: recover Office files
- Microsoft Support: view previous versions of Office files
- Google Drive Help: find or recover a file
What matters most
If the file matters, protect the choices before you chase the answer. A source label and a timestamp can save the recovery when two candidates look similar.