The most common way driver backups die is embarrassingly simple: they were saved on the same drive — sometimes the same partition — that got formatted five minutes later. The rule is physical, not digital. The backup must live somewhere the format cannot reach.
DISM will happily export to C:\DriversBackup on a PC you are about to wipe clean. The command cannot know your plans; it writes exactly where you point it. So choosing the destination is the entire skill of this step, and it is decided before the command runs, not after the format has eaten the evidence.
The destination scorecard. One test decides everything: does the format touch this location, or not.
If the format touches it, it is not a backup location. Windows Setup's format and delete steps act on partitions, and everything on a targeted partition goes — Windows, programs, and the neat DriversBackup folder sitting beside them. 'But I saved it in its own folder' changes nothing; folders are not armour.
State the test as a question before every job: if this partition is wiped in the next ten minutes, does my backup survive? If the answer needs a 'probably', the backup moves before the format starts.
A pen drive is the natural home for a driver export: 8 GB covers most machines, it is physically separate the moment you unplug it, and it plugs straight into the fresh install for the restore. Label it — DRIVERS, the PC model, and the date, written on a strip of tape, has saved more jobs than any app.
The eject step matters: a drive pulled while Windows is still flushing writes can keep the folder names and quietly lose the contents — a backup that passes a glance and fails the restore.
Many PCs carry a data drive or a D: partition that genuinely survives a Windows reinstall, and parking the backup there is legitimate — if you are formatting only the Windows partition and you verify that at the setup screen, row by row, before clicking anything.
This is the option with a human-shaped risk: technicians have wiped the 'safe' partition by selecting the wrong row in a hurry, and drive letters at the setup screen are not the letters Windows used. Read sizes and labels, not letters, and when a row is ambiguous, stop and identify it. The pen drive has no such failure mode, which is why it stays the default advice.
Uploading the zipped backup to Google Drive or OneDrive is excellent insurance and a poor bench tool. At restore time you need the files on something the half-built machine can read — and a machine with no network driver yet cannot fetch its own network driver from the cloud. That circle is exactly what the pen drive exists to break.
So use the cloud for what it is good at: copy number two, in a different building from copy number one. Pen drive lost, borrowed, or dead a month later, the cloud copy rebuilds it from any working PC. One copy is a hope; two copies in two places is a backup.
Four steps, in that order, every time. Every dead-backup story you will ever hear is one of these steps skipped, swapped, or done from memory instead of from the checklist — the format performed first and the verification attempted after.
Watch the export in the video and decide your destination before you run the command — that order is the lesson.
▶️ Watch Full VideoA pen drive — inexpensive, and the export fits easily. If no USB storage exists at all, upload the backup to cloud storage from the still-working Windows, then download it on any PC after the format and carry it back on whatever is available.
Yes. Any storage that stays disconnected or untouched during the format works equally well. The requirement is separation from the format, not the type or size of the device.
No. Setup formats or deletes only the partitions you select. The entire risk is human — choosing the wrong row quickly. Read the size and label of every row before every click and the risk disappears.