Presumably one has rolling backups. Something line N backup drives, and every day you back up to the next one in line. Then there are rolling monthly backups, etc. The idea is hopefully you notice the problem before all are corrupted.

With the way I understand ransomware works, merely connecting a backup to the system will corrupt it. This is where a write-enable switch really comes into play, it cannot corrupt the backup when you're trying to restore from it. And if you have online archival databases, they can't be corrupted, either. After all, how much of your company data really needs to be writable? Certainly not last week's payroll records.

With append-only incremental backup systems, enforced by hardware, it'll take a lot longer to need to recycle the backups and risk them. Tape backup already does this inherently.

Another thing you can do with a backup drive, is attempt to read it from a clean system that has never been connected to the network. If you can read it, it isn't corrupted. Then, have a process where once a week see if the backup can be read. Then your worst case is a week lost.