This is the protocol RestorePath Lab will use for future Windows data recovery software evaluations. It defines the dataset, safety boundaries, measurements, and evidence labels before any product result is published.
Safety boundary
- Use only disposable removable media, a disposable secondary drive, or a controlled virtual disk.
- Never delete test data from the Windows system drive or a device containing personal files.
- Keep the original dataset and its hashes on a separate known-good drive.
- Install the recovery program and save output on drives other than the test source.
- Do not simulate physical failure, repeatedly power-cycle unstable hardware, or use repair commands during a recovery measurement.
The reference dataset
Each run starts with 100 non-sensitive test files whose contents can be redistributed or generated for testing. The set includes DOCX, XLSX, PDF, TXT, JPG, PNG, MP4, and ZIP files in nested folders. It includes spaces, Unicode characters, long names, and duplicate extensions so that filename and folder reconstruction can be evaluated.
Before deletion, record each file's relative path, byte size, and SHA-256 hash. The untouched reference copy remains off the test source. Hash comparison later distinguishes an exact recovery from a file that merely opens.
Controlled scenarios
- Recent permanent deletion: delete the dataset from a disposable NTFS test volume, avoid further writes, and run the documented scan mode.
- Quick format: use a separate disposable volume, record the file system before and after, and never imply the result applies to every drive or format method.
- Media comparison: label HDD and SSD runs separately. Record delete-notification status and do not treat them as equivalent recovery conditions.
Any additional scenario needs its own written procedure and fresh disposable media. A product is not penalized for refusing an unsafe operation.
What the test record must include
| Record | Required detail |
|---|---|
| Environment | Windows edition and build, software version, test date, and relevant controller or connection. |
| Source | Media type, capacity, file system, scenario, and confirmation that the source is disposable. |
| Scan | Selected mode, elapsed time, interruptions, errors, and whether the scan session can be saved. |
| Discovery | Unique reference files found, correct paths and names, and representative previews—not the product's headline item count. |
| Recovery | Files saved to a separate destination, files that open, exact SHA-256 matches, partial files, and corrupt output. |
| Commercial limits | Free saving allowance, edition used, renewal term, current price check, and any affiliate relationship. |
How results are reported
- Exact recovery: the recovered file's SHA-256 hash matches the untouched reference.
- Usable but changed: the file opens and its expected content can be inspected, but its hash differs.
- Found but not verified: the entry or preview appears, but no usable output was recovered.
- Not found: the reference file was not identified in the selected scan.
We report counts and percentages for the reference dataset, not a universal star rating. Scan speed is recorded but not compared across different devices, interfaces, datasets, or test conditions.
Evidence and limitations
A published result should include a privacy-safe screenshot or table tied to the recorded run, the exact procedure, and a statement of what the run cannot prove. One successful NTFS deletion test does not establish performance on APFS, RAID, BitLocker, physically failing drives, or every SSD controller.
Vendor documentation can establish supported systems and features. Only the controlled run establishes what happened in that test. Learn how those labels fit the rest of the site in our full editorial method.
A reader can reproduce the core check
- Create a small non-sensitive dataset and retain an untouched copy.
- Record paths, sizes, and SHA-256 hashes.
- Use disposable media and document the exact loss scenario.
- Recover to a separate physical drive.
- Open the output and compare hashes before drawing a conclusion.
Official reference: Microsoft Windows File Recovery guidance for source/destination separation and mode selection.