The administrator who cannot go on leave
Single points of failure are usually people, not servers.
In many organisations the learning platform is understood by exactly one person. They know which report the finance team needs, which plugin breaks on upgrade, why the third category is named the way it is, and where the certificate template lives. None of this is written down, not through secrecy but because writing it down is never the most urgent thing on any given day.
The consequence is a person who cannot take two weeks off without leaving a set of instructions that would take two weeks to write. It is a bad outcome for them and a serious risk for the organisation, and it is almost always discovered at the worst possible moment.
The fix is unglamorous and incremental. Every time you solve something non-obvious, write four lines about it somewhere findable: what happened, what caused it, what you did, what to watch for. Not a manual, a log. After a year that log is the most valuable document in the department and it cost about ten minutes a week.
The second half of the fix is to make somebody else do a task occasionally while you watch. Reading instructions proves nothing. Performing the quarterly restore test, or generating the report, or processing a bulk enrolment, is what reveals the steps that were never captured because they live in muscle memory.
The log has a use beyond continuity that is worth naming, because it is the reason to keep it even if nobody else ever reads it. Written over a year, it is the only evidence you will have of what the job actually involved, and it is what allows an honest conversation about resourcing. An administrator who can say that this particular recurring fault consumed eleven afternoons is making an argument. One who says the work is heavy is expressing a feeling, and the two are received very differently by anybody holding a budget.