A register of interests that nobody reads is a filing exercise. The value is in the second half: this person declared that interest, this decision touched it, and here is what was done about it. Written down at the time it is unremarkable. Reconstructed afterwards, by somebody looking for it, it is the story.
Record the decisions as they are taken. The page matches them against the register by the party involved and tells you who had an interest, so a name that should have been absent from a vote is found before somebody else finds it.
An interest is not an accusation. Most declared interests are ordinary and harmless, and treating a declaration as an admission is the fastest way to stop people making them. A register works when declaring is routine and unremarkable, which means keeping it easy, keeping it current, and never using it as evidence of anything other than that somebody was open.
The second half is what matters. Everybody has a register. Very few have the other column: this decision touched that interest, and here is what was done. That is the column somebody will ask about, and it is the one that cannot be filled in afterwards without it looking exactly like what it would be.
The match is a prompt, not a verdict. Matching a decision to the register by the party involved catches the obvious cases and not the subtle ones: a supplier's parent company, a friend of a friend, an interest somebody never declared. It is a net with holes in it, which is better than no net, and it is not a substitute for anybody thinking.
An interest that has ended still belongs in the register. Deleting the row when somebody resigns a directorship removes the record that they held it during the period the decisions were taken in. Mark it ended, with the date. The register is a history, and a history with the inconvenient parts removed is not one.
It stays on this machine. A register of interests names people and their private affairs, and it is exactly the sort of document that should not be sitting on a service somebody signed up for. Export it and keep it where the governance papers live.
Anybody who takes or influences decisions: directors, trustees, senior staff, panel members, anybody procuring or awarding. The test is influence rather than seniority, and the people most often missed are the ones who write the recommendation somebody else signs.
Annually as a floor, and at the start of anything where it matters, such as a procurement or a recruitment panel. Each entry here carries its own review date so the question has an answer per person rather than once a year for everybody.
Anything that could reasonably be seen as capable of influencing the decision, which is a wider test than whether it actually would. The point is what a reasonable outsider would think on reading it, so the safe answer is to declare and let the register show you were open.
Record that you asked and what they said. A register that says "asked on this date, nil return" is a different document from one that is silent, and the difference matters when a conflict turns up later.
Sometimes. It depends on the decision, on your own rules, and on whether their presence would affect what others say, which it often does. The register offers both, along with not receiving the papers, because seeing the papers is participation.