which implies or can be extended by (according to what you mean by your terms) by “per installation type” which has several dimensions:
manual or not
dockerized/swarmed/k8sized/whatnotized
dev/production
type of machine [e.g. arm experimenter, openbsd adventurer]
being a newbee, lerner, dev, full-blown expert
how and where data is stored
knowing how and where it is stored
having reliable backups and a proven restoring path if the unspeakable happens
precisely when upgrading was stopped before updating
you name it
So maybe somebody can give a rough estimate, but without this type of detailed info, it’ll probably be very rough. But I’m not an Expert myself, just an humbled one when sometimes realizing my own knowledge gaps or being reawoken by mishaps, my own or otherwise, which might just happen randomly also.