Home › Common Problems › Repair vs Recovery

Repair, Recovery, or Rebuild

Last reviewed on May 11, 2026

When a file behaves badly, three different techniques are often described in the same breath: repair, recovery, and rebuild. They sound interchangeable but solve different problems, and applying the wrong one wastes time at best and destroys data at worst. This guide separates them and walks through the decision a user actually faces in the first ten minutes after noticing a file is broken.

Three different operations

Repair means fixing a file in place. The file still exists; the data is still there; some structural part of the file (a header, an index, a check value) is wrong, and repairing it means making the file valid again so its normal software can open it. Word's "Open and Repair", PDF readers that re-stream a corrupted PDF, and tools that rewrite a damaged ZIP central directory are all repair operations.

Recovery means pulling data out of a file (or a storage device) that can't be repaired. The original structure is too damaged to be made valid, so the goal shifts to extracting whatever readable content can be salvaged — text from a corrupt document, frames from a broken video, photos from an unmountable card. The output is usually a new file with the recovered content, not the original.

Rebuild means regenerating the file from its source. If the broken file is a PDF and the original Word document still exists, exporting a fresh PDF is faster and more reliable than any repair tool. If a video file is unrecoverable but the camera card still holds the raw clips, re-rendering produces a clean replacement.

Rebuild is almost always the best option when available. Recovery and repair are what you use when it isn't.

The decision in order

Run through these checks before reaching for any tool:

  1. Is there a working backup? If a recent backup of the file exists, restore from it. The "broken file" then becomes a curiosity, not a crisis.
  2. Is there a source the file was generated from? Exported PDFs, rendered videos, archived ZIPs, and compiled binaries can all be regenerated from their inputs. Regenerate rather than repair whenever you can.
  3. Was the file received from someone else? Ask for it again. A clean re-send beats any repair attempt, and the original may not have been corrupted on the sender's side.
  4. Does the application that normally opens the file have a built-in repair option? Word, Excel, Photoshop, video editors, and several archive tools all have one. These options are usually conservative and safe to try; they're the next step after backups and sources.
  5. Has the storage device the file lives on been behaving strangely? If yes, stop working with the file itself and move to device-level recovery — see below.
  6. If you've reached this step, repair tools and format-specific recovery tools are next. Work on a copy of the file, not the original.

Always work on a copy

Before any repair or recovery attempt, copy the file to a new location and leave the original untouched. Repair tools occasionally make a file worse — a partial-success result that overwrites the original is worse than a failure that leaves you the chance to try a different tool. The cost of the extra copy is a few seconds and a few megabytes; the cost of skipping it can be the whole file.

For very large files where space matters, a hash of the original (see the integrity verification guide) at least lets you tell whether a tool touched the file.

File-level damage versus device-level damage

One question matters more than any other and is the one most often skipped: is the problem with the file, or with the device the file lives on?

Signs that the problem is file-level only:

Signs that the problem is device-level (and that file-level repair is the wrong approach):

File-level repair on a device that's still failing will produce more broken files. The right next step in the device-level case is to clone the device — bit-for-bit, while it still responds — to a healthy disk, and work on the clone. For physically damaged drives, that means stopping immediately and consulting a professional data-recovery service; running consumer recovery software on a failing drive can permanently destroy data that a specialist could otherwise have rescued.

Worked example: a video file that won't seek

You have an MP4 from a phone that plays for a few seconds and then freezes. Walking through the decision:

  1. Backup? No.
  2. Source? The phone is still working and the clip is also in the camera roll — copy a fresh export from the phone. Done.

Same scenario, but the phone has been wiped. Now:

  1. Re-send? Not applicable.
  2. Application repair? A video player isn't a repair tool, but several format-specific tools can rewrite a broken MP4's index from the frame data; that's a repair operation.
  3. Device behaviour? Was the file received from a working drive? If so, file-level repair is appropriate. If the drive itself has been throwing errors, clone first.
  4. If repair fails? Recovery — try to extract the frames or audio that play correctly into a new file and accept the loss of the remainder.

The same flow applies to documents, archives, audio files, and project files. Substitute the corruption symptoms and format-specific tools from the relevant section of the site.

Common mistakes

When to stop and ask for help

Three signs that it's time to stop trying yourself: the device is making a sound it didn't make before, every tool you try produces a different error on the same file, or the file matters more than your tolerance for a bad outcome. Professional data-recovery services aren't cheap, but they have access to controlled environments, replacement hardware, and procedures that consumer tools can't match. The cheapest moment to involve them is before you've made things worse.

For everyday cases, the decision flow above keeps the steps in the right order and avoids the single biggest cause of unnecessary data loss: rushing.