A newsroom's data stewardship programme has stewards named for every dataset and almost no stewardship happening. What would you change?

Why named stewards do not produce stewardship

Naming someone in a spreadsheet assigns accountability without capacity, authority or feedback. The steward has a full-time role elsewhere, no time allocated, no tooling, and no consequence for inaction. The programme measures coverage — percentage of datasets with a named steward — which is achievable by editing a spreadsheet and correlates with nothing.

What to change

• Fewer stewards with real allocation. Steward the datasets that matter, with time explicitly budgeted, in preference to nominal coverage of everything.
• Authority to match the accountability. A steward who cannot refuse a bad change or require a fix is a contact point, not a steward.
• Work that arrives. Stewardship tasks routed to a queue — quality alerts, access requests, definition disputes, uncertain entity matches — rather than depending on someone remembering to check.
• Outcome measures, not coverage: time to resolve quality issues, proportion of definitions current, proportion of access requests handled within target.
• Domain placement. A steward inside the producing team knows the data; a central steward becomes a ticket router.

The framing that helps

Stewardship is an operational function, not a governance ceremony. It needs a queue, an SLA and staffing, exactly like on-call. Programmes that treat it as a documentation exercise produce documentation.