Main idea

A cross-cutting rule can perfectly well be written in the body of a Jira card, in a comment on a specification or in a roadmap line. There it is accurate, dated, findable by its author. It is also out of reach of everyone it concerns: the support agent who will answer the customer, the salesperson who will make a promise, the team that will build something else six months later.

Each artifact has an audience and a lifespan of its own. A ticket addresses those who execute and dies on delivery. A roadmap addresses those who want to know the horizon and is rewritten every quarter. A specification addresses those who build a given scope. Placing a rule valid for the whole company in one of them means giving it their audience and their life expectancy — that is, making it inoperative everywhere else.

The effect is asymmetrical and deceptive: the decision exists, it is written, you can find it if you know where to look. Nothing signals that it has disappeared, since it has not disappeared.

Why it matters

This gives a filing criterion that does not depend on the author's convenience: content goes where the audience and the duration match its own, not where you happened to be writing.

It also explains why an organization can have everything written down and know nothing: the question is not whether a decision is documented, but whether it is documented where those who depend on it will go and read it.

Nuances and limits

Bad filing is sometimes without consequence: a rule that really only concerns one team is well placed in that team's own tools, even if it has the appearance of a principle.

And a separate register does not guarantee visibility: a decision filed in the right place but never announced stays unknown to those who are not in the habit of going there.

Open questions

• How do you spot the structuring decisions already buried in a team's tickets and specifications, without rereading everything?