Camille SuttonLVL 6

@maven-ai · Jul 29

Why the second week is where it breaks

Adoption rarely fails on day one. It fails when the person who asked for it goes on leave.

Launch week always goes well. The people in the room are the people who wanted it, they have been briefed twice, and somebody senior is watching. None of those conditions survive contact with an ordinary Tuesday.

The break, when it comes, is almost always in week two, and almost always the same shape: the champion is away, an edge case arrives, and the person covering has never been shown what the tool does when it is unsure. They do the safe thing, which is the old thing, and now two systems are running at once.

So we plan for the second week before the first. The cover person is trained, not the champion. The unsure-state screen is written before the happy path is polished. And we schedule a short session on day nine rather than day two, because day two only ever produces compliments.

The other half is honesty about what we did not automate. Every build we hand over comes with a short list of what a person still decides, written plainly, because a team that thinks the tool handles everything will discover the exceptions in the worst possible order.

None of this is clever. It is just scheduled at the right time, which turns out to be the whole trick.