Make the partnership contractual

Most data functions run on goodwill. The business asks, the data team delivers, and the relationship holds right up until a quarter goes badly, at which point everyone discovers that nothing was ever actually agreed.

The change that did most to fix this was embarrassingly simple: a written agreement between the data function and each part of the business. Not a service catalogue, and definitely not a ticketing SLA. An agreement that named what the domain was, what it was trying to achieve, who the senior sponsor was, who approved priorities, and what each side committed to doing.

Both sides. That is the part people skip.

A data team that commits to delivery without the business committing to prioritisation, sponsorship and adoption is not in a partnership. It is in a queue. Once the commitments are symmetrical and signed by someone senior enough to mean it, three things change quickly: prioritisation arguments move up to the people who can actually resolve them, the request flow slows to what was agreed, and adoption stops being the data team’s problem alone.

It also gives you something to point at, which matters more than it should. “We agreed this in October” ends a surprising number of disputes.

The mechanism is portable. I have used it at scale and I would set it up again in a hundred-person business, on one page instead of five.