3 ideas
Treat data the way a factory treats physical product: it moves through a lifecycle of creation, storage, usage, and disposal, and each stage requires defined controls and accountable owners. Managing data quality means inspecting and correcting it at the points where it is produced or transformed, not after it has already polluted downstream reports.
Accountability for data should split into two distinct roles: the data owner who holds decision rights and answers for outcomes, and the data steward who executes day-to-day definition, quality monitoring, and issue resolution. Conflating these into one person collapses governance because the authority to set rules gets mixed with the labor of enforcing them.
A data governance program defined only through policies, committees, and org charts will not change behavior; it must be attached to concrete operational deliverables like business glossaries, data quality rules, and metadata that people use in their actual work. Governance becomes real only when it produces artifacts that practitioners depend on rather than documents that sit unread.