Data recovery results depend on the storage device, file system, loss event, overwrite history, encryption, physical condition, and software version. A successful result on one healthy device does not prove that every deleted file can be recovered from every device.
Our testing process is designed to answer narrower and more useful questions:
- What device, file system, software version, and loss scenario did we test?
- What files did the software detect and actually export?
- Did the recovered bytes match the original files?
- Could the recovered files be read or opened?
- Were the original file names and folders preserved?
- Did the scan create duplicates or unrelated outputs?
- What does the test not prove?
This page explains our method, summarizes the results we can currently support, and separates completed tests from planned or excluded scenarios.
Current Test Coverage
As of November 19, 2026, our documented test program covers:
- permanent deletion from a physical FAT32 USB flash drive;
- ordinary deletion and Recycle Bin recovery on a FAT32 volume;
- quick-format recovery from a FAT32 USB flash drive;
- quick-format recovery from a physical FAT32 SD card;
- data recovery from a physical external HDD;
- a controlled RAW or inaccessible FAT32 volume;
- a controlled lost MBR partition;
- a separate formatted FAT32 volume test.
The tests used different versions of Drecov and different test environments. We do not combine them into one universal recovery rate.
Published Test Summary
| Scenario | Environment and version | Baseline result | Names and paths | Overall record |
|---|---|---|---|---|
| S01 Permanent deletion | Physical FAT32 USB; Windows 10 and Windows 11; Drecov 2.1.3.0 | Three runs, each with 1,470/1,470 exact SHA-256 matches and 1,470/1,470 read checks | 1,470/1,470 original names retained in each run | Complete |
| S02 Ordinary deletion and Recycle Bin supplement | FAT32 volume; two independent 1,400-file groups | 1,400/1,400 exact matches in each group | Both groups retained file names and test-directory paths | Complete |
| S03 USB quick format | Physical FAT32 USB; Drecov 2.1.3.0 | 1,380/1,380 exact matches and read checks | 1,380/1,380 original names retained | Complete |
| S04 SD card quick format | Physical FAT32 SD card; Windows 10 and Windows 11; Drecov 2.1.3.0 | 1,786/1,786 exact matches | 1,786/1,786 original names retained; 1,786/1,786 relative paths retained | Complete |
| S05 External hard drive | Physical external HDD; Windows 10 and Windows 11; Drecov 2.1.3.0 | 1,600/1,600 exact matches | 1,600/1,600 original names retained | Complete |
| S06 RAW logical damage | Controlled FAT32 fixed VHD on a local SATA host; Windows 10 and Windows 11; Drecov 2.1.3.0 | 1,786/1,786 exact matches | 1,786/1,786 original names retained; 1,786/1,786 relative paths retained | Complete |
| S07 Lost partition | Controlled 1 GiB FAT32 MBR VHD; Windows 10 and Windows 11; Drecov 2.1.3.0 | 1,786/1,786 exact matches | 1,786/1,786 file names, relative paths, and complete directory tree retained | Complete |
Test 1 Permanent Deletion from a Physical USB Drive
We completed three independent permanent-deletion tests on the same dedicated Netac USB flash drive. The source was a physical USB device with a FAT32 volume. The recovery destination was an NTFS volume on a different physical SATA disk.
Each run used 1,470 files:
- 1,461 newly generated synthetic samples;
- 9 samples created to revisit earlier WAV and BMP findings.
Before deletion, every source file was checked against an independent SHA-256 manifest. The test directory was then permanently deleted without using the Recycle Bin. Each scan and export used its own run identifier, baseline, output directory, and verification record.
| Run | Exact content matches | Read checks | Original names | Scan completion interval |
|---|---|---|---|---|
| S01-USB-R1 | 1,470/1,470 | 1,470/1,470 | 1,470/1,470 | 1,658.588–1,659.768 seconds |
| S01-USB-R2 | 1,470/1,470 | 1,470/1,470 | 1,470/1,470 | 1,658.890–1,660.021 seconds |
| S01-USB-R3 | 1,470/1,470 | 1,470/1,470 | 1,470/1,470 | 1,658.920–1,659.110 seconds |
All 1,470 baseline files matched their original content in all three runs. The 20-second WAV sample that had failed a historical checksum test matched the known original checksum in the new tests and completed silent playback.
The current USB differs from the approximately 29 GB medium used in the historical run. Later successful content recovery therefore does not prove that every historical condition has been fixed.
Test 2 Ordinary Deletion and Recycle Bin Recovery
An earlier supplemental test used two independent 1,400-file groups on a FAT32 source volume.
In the first group, the files remained in the Windows Recycle Bin. Drecov exported all 1,400 files, and all 1,400 matched their original SHA-256 checksums. The directory paths within the test folder were retained.
In the second group, only the test directory was permanently removed from the Recycle Bin. This was not a global “Empty Recycle Bin” operation. Drecov recovered all 1,400 baseline files with exact SHA-256 matches and retained their file names and test-directory paths.
This result is published as a FAT32 supplement documenting recovery after ordinary deletion and permanent removal of a test directory from the Recycle Bin.
Test 3 Quick Format of a Physical USB Drive
The S03 test used an independent 1,380-file dataset on a FAT32 USB drive. Every source checksum was verified before a real FAT32-to-FAT32 quick format. Drecov then scanned the formatted source and exported the results to another physical disk.
| Result item | Recorded result |
|---|---|
| Baseline files | 1,380 |
| Exact SHA-256 matches | 1,380/1,380 |
| Full-read checks passed | 1,380/1,380 |
| Original names matched | 1,380/1,380 |
| Actual exported files | 1,380 |
The known baseline content was recovered exactly. The entire medium was overwritten before the quick-format scenario, allowing additional exported results to be used in a strict false-positive-rate calculation.
Test 4 Quick Format of a Physical SD Card
The S04 test used a physical 31,266,439,168-byte SD card connected through a USB card reader. The card was FAT32, MBR, healthy, and not a system disk. The recovery destination was a healthy NTFS volume on a different physical SATA disk.
The dataset contained 1,786 files totaling 26,843,545,600 bytes across 134 extensions. It combined 1,648 collected/generated samples with 138 synthetic samples. Each baseline file had a unique SHA-256 value.
After verifying the source data, we performed a real FAT32-to-FAT32 quick format. Drecov 2.1.3.0 scanned the SD card and exported the selected results.
| Result item | Recorded result |
|---|---|
| Baseline files | 1,786 |
| Exact content matches | 1,786/1,786 |
| Original names matched | 1,786/1,786 |
| Relative paths retained | 1,786/1,786 |
| Format-specific read checks passed | 1,666 |
| Read checks unavailable in the test toolset | 120 |
| Observed scan completion interval | 1,374.742–1,376.979 seconds |
The test verified all baseline contents by checksum.
Representative PNG, MP4, and PDF files were previewed in Drecov. Representative DOCX, XLSX, PPTX, PDF, PNG, JPG, TXT, ZIP, WAV, and MP4 outputs were also checked after recovery. These representative checks do not mean that all 134 extensions were manually opened in their native applications.
Test 5 Recovery from a Physical External Hard Drive
S05 used a physical external HDD and Drecov 2.1.3.0 in Windows 10 and Windows 11 environments. The test dataset contained 1,600 files. After recovery, the exported files were compared with an independent baseline manifest created before the test.
| Result item | Recorded result |
|---|---|
| Baseline files | 1,600 |
| Exact content matches | 1,600/1,600 |
| Original names matched | 1,600/1,600 |
| Overall record | Complete |
Why a Hard Drive May Offer More Favorable Recovery Conditions
A traditional hard disk drive usually does not behave like a TRIM-enabled SSD or some flash-based devices that may actively clear deleted blocks. If the original sectors have not been overwritten and the drive remains readable, recovery software may still locate data through file-system records or file signatures. Under the same logical-deletion conditions, a healthy HDD that has not received further writes may therefore offer more favorable recovery conditions than flash storage affected by TRIM or internal garbage collection.
This does not mean that an HDD always has a higher recovery rate, and it is not a recovery guarantee. Recovery may still be limited by:
- new writes that overwrite the original sectors after deletion;
- heavy fragmentation after file-system indexes have been lost or damaged;
- bad sectors, head problems, abnormal sounds, overheating, or repeated disconnections;
- unstable USB bridges, cables, or power delivery;
- damaged partition tables, file-system metadata, or encryption information;
- internal behavior of shingled magnetic recording (SMR) drives.
How We Determine Whether External-HDD Recovery Succeeded
The external-HDD test follows the same validation principles as our other scenarios. We create a baseline manifest outside the source, then record detection, export, SHA-256 content matching, and original-name preservation separately. A file counts as an exact content match only when the exported bytes match the pre-test baseline. A scan result, visible file name, or identical file size alone does not prove successful recovery.
The recorded result was 1,600/1,600 exact baseline-content matches and 1,600/1,600 original-name matches. The test is therefore recorded as Complete within the requirements that were defined and documented for this run.
This conclusion applies only to this controlled test and must not be treated as a fixed recovery rate for every hard drive. The public record does not disclose the drive capacity, file system, exact loss action, scan time, relative-path retention, or device-health measurements, so those items are outside the scope of the published conclusion. If a hard drive clicks, overheats, disconnects repeatedly, or has severe bad sectors, repeated scanning should be stopped and professional recovery should be considered.
Test 6 RAW or Inaccessible Volume Recovery
The latest S06 retest used a dedicated 1 GiB fixed VHD hosted on the test computer’s local SATA disk. The source was created as a normal FAT32 volume and populated with the same 1,786-file, 134-extension baseline.
To create a controlled logical failure, we zeroed the primary and backup FAT32 boot sectors inside the offline test VHD. We did not damage a physical disk or modify the physical host’s partition table. After the change, Windows reported the file system as Unknown and could not read the root directory.
Drecov 2.1.3.0 scanned the inaccessible logical volume and exported the results to a separate folder. The source and output were on different logical volumes but shared the same physical host disk, which is an important limitation.
| Result item | Recorded result |
|---|---|
| Baseline files | 1,786 |
| Exact content matches | 1,786/1,786 |
| Original names matched | 1,786/1,786 |
| Relative paths retained | 1,786/1,786 |
| Format-specific read checks passed | 1,666 |
| Read checks unavailable in the test toolset | 120 |
| Observed scan completion interval | 11.727–14.192 seconds |
After recovery, the source image was restored to its original FAT32 state and all 1,786 source files were verified again. The RAW image checksum was unchanged through the scan and export process in this test.
This was a controlled logical RAW scenario on a VHD. It does not prove recovery from a physically failing device, another file system, or every real-world RAW condition.
Test 7 Lost MBR Partition Recovery
The S07 test used another dedicated 1 GiB FAT32 fixed VHD with an MBR partition. The same 1,786-file baseline was written and verified. We then removed only the 64-byte MBR partition-table region from the offline test image. The FAT32 boot sector and file data remained in place, but Windows no longer showed the original partition or drive letter.
Drecov identified the lost partition and exported the selected files. No physical disk partition was deleted, and the test did not ask Drecov to write a partition table back to the source.
| Result item | Recorded result |
|---|---|
| Baseline files | 1,786 |
| Exact content matches | 1,786/1,786 |
| Original names matched | 1,786/1,786 |
| Relative paths retained | 1,786/1,786 |
| Complete source tree retained | 1,786/1,786 |
| Format-specific read checks passed | 1,666 |
| Read checks unavailable in the test toolset | 120 |
| Observed scan completion interval | 7.234–9.656 seconds |
The source image checksum remained unchanged during scanning and export. After the test, the original image was restored and all 1,786 source files were verified.
The source VHD and recovery destination were outside each other but shared the same physical SATA host disk. The test covered one FAT32 MBR partition and did not cover GPT, complex multi-partition damage, physical disk failure, or partition-table repair.
Supporting Formatted FAT32 Volume Test
A separate formatted FAT32 volume was evaluated with a larger file-instance baseline.
| Result item | Recorded result |
|---|---|
| Baseline test files | 1,980 |
| Exact SHA-256 matches | 1,980/1,980 |
| Unique baseline contents matched | 1,116/1,116 |
| Original names matched | 1,971/1,980 |
All baseline file instances were found with byte-identical content. The historical report did not record all environment fields needed for direct comparison with the later S-series tests, including a standardized software-version record. We therefore treat it as supporting evidence for that test, not as a general recovery rate.
Why We Repeat Tests
An earlier FAT32 USB test illustrates why file presence and file size are not enough.
The historical run used 1,380 files. Drecov found all 1,380 expected files, but only 1,370 matched their original SHA-256 checksums. One WAV file had the same byte length as the original but different content. Nine BMP and WAV names were also incomplete.
The later S01 physical-USB retest reconstructed the known 20-second WAV baseline and recovered byte-identical content in all three runs. We therefore record content recovery and name preservation as separate results rather than making a broad statement that every historical issue has been resolved.
How We Build a Test Dataset
We use synthetic or controlled files rather than private user data. Depending on the test, the dataset can include documents, images, audio, video, archives, email files, structured data, fonts, design files, disk images, and boundary cases.
Before deletion, formatting, or logical damage, we create a manifest outside the source. The manifest records where applicable:
- run identifier;
- original relative path;
- file name and extension;
- exact byte length;
- SHA-256 checksum;
- file category;
- source and destination information.
After copying the dataset to the source, we calculate the checksums again. The loss event begins only after the source files match the independent baseline.
How We Create a Loss Scenario
We record the exact action used to create the scenario.
- Permanent deletion: Remove only the authorized test directory without sending it to the Recycle Bin.
- Recycle Bin: Record whether files remain in the Bin, a test folder is permanently removed, or the Bin is globally emptied.
- Quick format: Record source and target file systems and confirm that the operation was quick rather than full format.
- RAW or inaccessible volume: Modify only a controlled offline image and record the exact sectors changed.
- Lost partition: Remove only the known partition entry from an authorized test image while preserving rollback data.
Destructive actions are performed only on identified media or images that are approved for erasure. We do not use real user partitions as test targets.
How We Verify Recovery
We do not treat a scan count or a “scan complete” message as proof of recovery.
Detected
The software displayed a possible file, folder, volume, or partition record. Detection does not prove that the content can be exported or used.
Previewed
The software displayed recognizable content before export. Preview evidence is recorded separately because not every file type supports preview.
Restored
The software exported a file to the selected destination. An exported file is not automatically complete.
Verified
The restored file matched the original SHA-256 checksum. In a controlled test, this is our strongest evidence that the recovered bytes are identical to the baseline.
Read Checked
Where tools are available, recovered files receive format-specific checks such as:
- full image decoding;
- complete audio or video stream decoding;
- ZIP or GZIP CRC validation;
- Office and OLE container parsing;
- PDF page and text extraction;
- database or table readback;
- structured-text parsing.
If a format-specific reader is unavailable, a byte-identical checksum can still be recorded as Verified, but not as Opened in its native application.
Opened
A representative recovered file was opened in a relevant application and recognizable content was observed. Representative opening does not mean that every file in the dataset received a full native-application check.
Name and Path Checked
Original names, relative paths, and the complete source tree are separate metrics. Byte-identical content can be recovered even when the original name or folder structure is missing.
Extra Outputs Recorded Separately
Recovery software may export duplicates, reconstructed files, file-system metadata, or historical content. We do not add these outputs to the successful baseline count. On media with a prior history, they also cannot automatically be labeled as false positives.
What Complete Partial and Failed Mean
| Status | Meaning in our reports |
|---|---|
| Complete | All requirements defined for the test, including content and required organization or behavior, passed. |
| Partial | The core content result passed, but another required item such as names, folders, environment isolation, preview coverage, or output quality remained incomplete. |
| Failed | Required baseline content could not be detected, exported, verified, or read as defined by the test. |
| Not tested | The required device, environment, or procedure was not available or was not executed. |
These labels apply only to the named test. “Complete” never means that recovery is guaranteed for every device.
Safety Rules During Testing
Our controlled tests follow these rules:
- Verify the source identity before a destructive action.
- Use only authorized test media, images, and non-private samples.
- Keep the baseline manifest outside the source.
- Do not install recovery software on the affected source.
- Do not intentionally save recovered files back to the source.
- Keep detection, preview, export, opening, and checksum verification as separate states.
- Preserve rollback data for controlled RAW and partition tests.
- Do not run recovered executables, scripts, macros, drivers, or mobile packages as part of validation.
- Do not present a logical recovery test as evidence of physical-device repair.
- Do not test password bypass or unauthorized access.
Not every test used a hardware write blocker or a separate physical destination. Those limitations are disclosed with the relevant result rather than assumed away.
File Types in the Expanded Dataset
The expanded 1,786-file dataset covers 134 extensions. It includes examples from these groups:
- documents and office files;
- PDF, text, HTML, XML, CSV, and structured datasets;
- common and modern image formats;
- RAW camera file examples;
- audio and video containers;
- ZIP, GZIP, RAR, 7z, JAR, and other archives or containers;
- email and mailbox formats;
- design, CAD, font, project, and disk-image files;
- executables, libraries, scripts, and system-file samples used only for byte-level validation.
Some formats were verified by checksum but did not have an available format-specific reader or native application in the test environment. We do not convert those cases into an “opened successfully” claim.
How We Review Other Data Recovery Tools
When we publish a competitor comparison, we separate hands-on observations from official documentation. A product page or manual is not described as a laboratory result.
Where practical, comparison tests should use the same source, dataset, loss event, destination rule, and validation method. Pricing, free limits, compatibility, and vendor claims are checked against current official sources at the time of review.
PandaOffice develops and sells Drecov. That commercial relationship is disclosed. Payment cannot change a recorded test result or remove a relevant limitation from an editorial conclusion.
When We Retest
We schedule a retest when:
- Drecov or a reviewed tool receives a material version update;
- Windows or macOS changes storage behavior or permissions;
- a test reveals content corruption, naming loss, path loss, or unexpected output;
- a new device, file system, or loss scenario is added;
- a reader or reviewer reports a reproducible problem;
- an article makes a claim broader than the available evidence.
Where possible, a defect-focused retest reuses the original baseline samples. A new sample set can add evidence, but it does not automatically close an issue found with an older dataset.
How to Read Our Results
Published numbers describe a recorded environment, not a universal product guarantee. Recovery can fail when data has been overwritten, SSD TRIM has cleared deleted blocks, encryption keys are unavailable, a device is not detected, or physical damage prevents stable reading.
If a drive disconnects repeatedly, becomes unusually hot, makes abnormal mechanical sounds, reports an implausible capacity, or contains irreplaceable data, repeated software scans may be the wrong next step. Stop using the device and consider a qualified professional recovery service.
Corrections and Questions
If a testing statement needs clarification, contact support@pandaoffice.com and include the page URL, Drecov version, operating system, device type, file system, and the exact result you observed. Do not send private files, passwords, encryption keys, license keys, or screenshots containing sensitive information.
Related Pages
Frequently Asked Questions
Do you test every data recovery scenario?
No. Planned work is not presented as a completed test.
Does a scan result mean that a file was recovered?
No. A scan result only shows that the software detected a possible record. In a controlled test, we count content as verified only after export and comparison with a known baseline where one exists.
Why can recovery results differ between devices?
Storage technology, file system, deletion history, overwrite activity, TRIM, encryption, fragmentation, physical condition, connection quality, and software version can all change the result.
Do you publish a universal Drecov recovery rate?
No. A percentage from one test cannot represent every device or real-world loss event. We publish the result together with its scope and limitations.
Can Drecov recover overwritten files?
No recovery software can reliably recreate original bytes after they have been overwritten. Backups, previous versions, exports, or other copies may be the only remaining options.
Can software recover a physically damaged drive?
Logical recovery software requires stable access to the device. Clicking, overheating, intermittent detection, electrical damage, or severe read instability may require professional recovery rather than repeated scans.
Do you test competitor software?
We may perform version-specific hands-on checks or review official documentation. Each article should state which type of evidence supports its conclusions.
When is a test-based article updated?
We update a test-based article when the product version, operating system, test evidence, limitation, or conclusion changes materially. A cosmetic edit alone should not be presented as a new test.








