Already shipping, and still taking orders
Seven product teams on one platform. Digital, delivering, and organized around projects that were specified elsewhere and handed down.
That is the harder version of this problem. A company that has never shipped digitally knows it needs to change. A company that ships every week has evidence that the current way works, and the argument for changing it has to be made against a backdrop of things visibly getting done.
The change is agreed above the teams, or it is not agreed
The shift is from project-based delivery to a product-led model with team ownership, continuous discovery and OKRs.
I partner with the Head of Product, the Director Product Management and the CPTO on OKRs. That is where this has to be settled before any team is asked to work differently, because a team told to own outcomes while its roadmap still arrives from elsewhere has been given responsibility without authority, and will correctly ignore the instruction.
Prioritization with a written basis
Nothing enters the backlog on a hunch.
The discipline runs off the repository: a hypothesis has to be traceable to evidence somebody can open. A link to the study, and not a reference to research in general. That single rule does more for decision quality than any framework, because it makes the absence of evidence visible at the moment the decision is being made rather than six months later.
It also changes who can object. When the basis is written down, a junior designer can point at the gap, and the gap is the argument.
Documentation that stays current
A repository nobody updates is worse than no repository, because people trust it and it is wrong.
Ownership of each document is named, and the review cadence is part of the role description that names it. Documentation maintained out of goodwill decays at exactly the speed the organization grows.
Decisions
Settle it above the teams before asking anyone to work differently. Planned: roll the model out one team at a time, starting with the team most ready for it and letting the rest follow a good example. Argued: starting with the willing team proves nothing to the unwilling one, because what is being asked for is not effort. It is authority over a roadmap that arrives from somewhere else, and no team can grant itself that. Beginning low teaches everybody that the new model is something you opt into. Happened: it was settled in the OKRs, with all three of them signed up, before any team was asked to change a single ceremony.
Make one evidence rule the condition of entry. Planned: a weighted prioritization model, scored, applied to the backlog that already existed. Argued: a score turns a missing input into a number, and a number reads as an answer. One rule does more than a model: an item carries a link to the study behind it or it waits outside the backlog. Happened: that rule, and a side effect worth more than the rule. A gap in the evidence is now visible to anybody at any level, which changes who is able to object.
Where it stands
This is in progress and worth saying plainly: seven teams do not change how they decide in a quarter. What exists is the agreement at the top, the evidence rule in the backlog, and the discovery infrastructure that makes the rule possible to follow.
What does not exist yet is the number. I would measure the share of roadmap items traceable to evidence, and I did not start measuring it early enough.