Data ownership must resolve a real problem

Ownership requires a clear definition, an operating process and a route for users to raise issues.

I begin with a measure or dataset that affects a consequential decision. Working through one concrete problem reveals the responsibilities a broader governance model must support. It also makes the value of that model visible to the business.

Ownership models fail for a familiar reason. They are designed in the abstract, with domains, roles and responsibilities drawn up before anyone has tested whether the named owners have the authority, time or knowledge to act. Senior people accept the title because declining it seems unhelpful, then delegate it to someone without decision rights. When a quality problem appears, users still do not know whom to contact, and the governance document offers a name rather than a resolution. The model exists on paper while the underlying frustrations continue.

My starting point is a single measure that matters to a leadership decision and is currently disputed. I bring together the people who produce it, the people who use it and the people who report it, and ask them to agree a definition, a source and a set of quality checks. We then identify who can change the definition when the business changes, and how users should raise a problem. Writing down what that person actually needs in order to act tells us more about ownership than any generic framework.

Consider a business in which the customer count reported to the board differs from the one used by sales and the one held in the finance system. Each team has a reasonable definition: active accounts, contracted accounts, billed accounts. The dispute consumes time in every review and undermines confidence in the figures. Agreeing one primary definition, documenting the others as legitimate variants and naming an owner with the authority to maintain them removes a recurring argument. It also creates a working example of what ownership means in this organisation.

The counter-argument is that starting with one problem is too slow when the organisation holds a large estate of data that needs governance. I understand the impatience, and regulatory obligations sometimes require broad coverage quickly. Even then, I prefer to establish a minimal baseline across the estate while proving the full model on a few critical measures. A comprehensive framework imposed before anyone has seen it work tends to generate compliance activity without improving the data. Demonstrated value gives the wider rollout a better chance of being taken seriously.

This asks business leaders to accept ownership as a real responsibility, with time and authority attached, rather than an honorary title. It asks the data function to support owners with tools and expertise instead of taking the responsibility away from them. And it asks the leadership team to resolve definitional disputes when owners cannot, because some disagreements reflect genuinely different business interests. The board can help by asking which consequential measures now have accountable owners, and what changed as a result, rather than how many policies have been approved.

Starting with a real problem also builds the credibility that governance needs. Business leaders who see ownership resolve a recurring frustration are far more willing to extend the model to the next domain. Governance then grows because it is useful, not because it is mandated.