There is a particular kind of satisfaction in finishing a set of household records. The folder is assembled, the list is written, the documents are located and labelled. It feels like a completed task, and it is treated as one.
It is not a completed task. It is the first instance of a recurring one, and the difference between those two framings determines whether the work is still worth anything in five years.
What decays, and how fast
Records decay unevenly, which is what makes the decay hard to notice. Some elements are stable for decades: the existence of a will, the identity of your children, the fact that you own the house. Others are stable for a few years: which bank, which insurer, which adviser, which address. Others change constantly: balances, values, the specific state of any account.
A record written once mixes all three categories on the same page, and the page ages at the rate of its fastest-moving element. Within a year or two it contains a mixture of true, stale, and actively misleading statements, with nothing distinguishing them.
Actively misleading is the important category. A missing entry causes a search. A wrong entry causes a search in the wrong place, followed by a period of confident error. The record that names a bank the household left in 2023 is worse than one that named no bank at all, because it will be believed.
Why the first version is the weakest
Not only because it ages, but because the first attempt at any record is made without knowing what it will be used for. The household guesses at what an executor will need, and guesses reasonably, and gets some of it wrong — too much detail in places that do not matter, none at all in places that do.
A review corrects that, because between the first version and the second something usually happens that tests it. A relative dies and the family sees an estate administered up close. Somebody is ill. A parent moves into care. Each of those teaches the household something specific about what the record failed to contain, and a review is the only mechanism that captures the lesson.
What a review actually is
Shorter than the first pass, and different in kind. The first pass is compilation. A review is three questions, applied to what already exists.
Is anything here no longer true? Institutions merge, accounts close, people move, executors become unsuitable or unwilling, relationships change. The nominated beneficiary on a pension outlives its accuracy more often than any other single item in a household's records.
Is anything missing that now exists? New account, new policy, new property, new obligation, new person who matters.
Would this still work if it were needed tomorrow, read by the person it is addressed to? Not by you — by them. The test is not whether the record is accurate. It is whether it is usable by somebody who cannot ask you a follow-up question.
The document is the artefact. The review is the product.
Annual, and why the date matters more than the interval
The precise interval is not important; the existence of a fixed one is. Records reviewed when someone remembers are reviewed once. Records reviewed on a date are reviewed indefinitely, because the date does the remembering.
Attaching it to something already annual works better than a standalone reminder — the same week as a tax return, an insurance renewal, a birthday. What matters is that the trigger is external, because the internal one has already been tested by every household that assembled a folder once and has not opened it since.