Data governance is about decisions, not data.
Metadata gaps, lineage tooling, framework maturity — that's how governance gets pitched to construction executives, and it's why they stop listening. There is a better opening question: who decides when this number changes?
Every capital-project organization has lived the same meeting. Two dashboards show two different values for the same KPI. The programme director asks which one is right, and the honest answer is that both are — each according to its own source, its own cut-off date, and its own unwritten definition of what counts. The meeting moves on. The trust does not come back.
When the data team proposes governance as the fix, the pitch usually leads with instruments: a metadata catalog, lineage tooling, a maturity framework, a council. All of it sounds operational — a cost, not a capability — and the sponsorship evaporates before the first workshop.
Data governance is not about data. It is about who controls decisions.
Reframe it. The business already cares about revenue growth, cost control, risk reduction, faster decisions and accountability. Governance is the mechanism that makes those possible, and the bridge is decision authority: clear metric ownership, definition rights, controlled changes, escalation paths. Not "we need lineage tooling" but "when the forecast completion date moves, who is allowed to move it, and who gets told?"
The questions that land
In practice, four questions open the conversation better than any framework slide. Who decides when this KPI changes? What risk does this number expose us to if it is wrong? What revenue or funding decision does it protect? How do we avoid two versions of the same number in front of the board?
Each question terminates in a governance artifact — an owner, a definition, a change-control rule, a single source — but none of them starts there. That is the entire trick. Decision authority is the headline; data management (quality, lineage, metadata, change control) is the machinery that makes decisions reliable; trusted outputs (consistent KPIs, dependable dashboards, scalable AI) are the payoff.
Why this matters more in construction
Capital projects run on numbers that move: forecasts, baselines, variations, claims. Every one of those movements is contested — commercially, contractually, sometimes legally. A governed date is not bureaucratic nicety; it is the difference between a defensible extension-of-time claim and an expensive argument. The industry does not have a data problem so much as a decision-rights problem that shows up in the data.
And it compounds. Every AI use case now arriving on site — predictive delay models, generative reporting copilots, reasoning agents evaluating procurement scenarios — inherits the governance of the data underneath it. Ungoverned inputs do not average out; they amplify. The organizations getting value from AI in 2026 are, with few exceptions, the ones that quietly fixed decision authority over their data first.
The simple translation
If the executive summary has to fit on one line, use this: data governance is clear rules for how business decisions are defined and changed. Start there, and the catalog, the lineage and the council all get funded — because now they are protecting decisions the business already values.