Five pages to send your board

Every other route through the Cabinet is built for somebody who will sit down and read. You are going to forward this one, and then lose control of what happens to it: opened on a phone in a taxi, skimmed at the end of a long agenda, read closely by exactly one person, who will be the one who asks you the hard question in the room.

So the selection rule is not which pages teach the most, but which survive being read cold by somebody with no context, no vocabulary and no patience, and still leave them able to vote. That rule is also why the route is five pages rather than thirteen. A board pre-read of thirteen pages is one nobody opens, and the consequence of sending it is directors who arrive having read your covering email and nothing else. The shortness is the design, not a compromise in it.

The first page describes the whole machine once, in ordinary words, and it comes first because every objection raised later in the meeting is an objection about one of its parts. A director who has read it can follow the rest of the pack without asking anyone to define a term, which is the difference between a discussion and a tutorial.

The middle three are the questions a board actually puts to a proposal of this kind, in the order it puts them: whether to buy this at all rather than build it, what the agreement has to say if the answer is buy, and what the bill does in the second year once the pilot has become a standing cost. Read as a run they are a decision paper. Read singly each is still complete, which matters, because a director will read the one that touches their own committee and skip the rest.

The last page is the shortest and it is last on purpose. It is the one a non-executive director asks for by name, usually phrased as a question about who is accountable when the thing gets something wrong, and it answers with a mechanism rather than a policy. Placed earlier it reads as a small technical detail. Placed here it reads as the control the whole arrangement rests on.

One caution about the second page. A chief executive who forwards this route is forwarding a comparison of buying with building to their own chief technology officer, and the page was written knowing that. It does not argue that your engineers could not build this. It sets out the parts of the job that are invisible from outside, so a build proposal can be scoped against the whole of it rather than the interesting quarter. A team that has already scoped the whole of it will recognise its own plan there and say so.

If your board has less time than this, send the first page and the fourth. That is about twenty minutes between them, and it leaves your directors able to ask what the agent is permitted to do without a person, and which line in the bill has no ceiling on it. Those are the two questions that change a vote.

What this route deliberately leaves out is the implementation. A board decides whether and on what terms; it does not run a rollout, and a pack that covers both reads as a project plan looking for approval. An account of the first ninety days is written for whoever will do it and belongs in your management reporting rather than in the papers. The eleven-part vocabulary is also out, and what that costs is a director who wants to go one question deeper than the pack allows. It sits in the anatomy, where the one director who wants that detail can find it without it being posted to eleven others who do not.

This route is for the room. If you are the person who will answer to that room every quarter from here on, the longer board route is written for you rather than for them, and it goes where a pack cannot.