I think if the patches are bundled together now - you basically have to treat them as one larger patch. In other words, nothing changes except any time you find you did one and it breaks something, you roll the whole collection back until it can be rectified.
To a certain degree, it's already that way.
This month, I have a customer with a Hyper-V cluster which one of the six patches screwed up iSCSI while backing up. And a customer with a Terminal Server which one of the six patches screwed up Terminal Services. And a customer with Exchange that one of the six patches broke Backup Exec being able to see inside the database to restore individual files.
Only in the case of the TS problem has it been tracked down to a single patch - by Microsoft. The other two batches, nobody knows which one is at fault. These are production machines and I don't have time to reapply patches one by one to help Microsoft isolate which one is bad. So yeah, after this unusually brutal month I'm okay with cumulative patches. I'm having to roll back batches anyway.