Data governance covers several related concerns:
- Ownership. Every important data set and metric has someone accountable for it, and often a steward who looks after it day to day.
- Definitions. Business terms and metrics are defined once, in a shared place, so that the same word means the same thing everywhere.
- Quality. Expectations for accuracy, completeness and timeliness are set and checked.
- Access and privacy. Rules determine who can see what, with particular care for personal and sensitive data.
- Change control. Changes to definitions, rules and data structures are reviewed, recorded and communicated.
- Lineage. It is possible to show where data came from and how a figure was produced.
In pharma commercial analytics, governance has a practical purpose: making sure that the brand team, finance, analytics and leadership get the same answer to the same question, and that the company can show how its figures were produced. Patient-level data, which pharma companies typically receive in de-identified form, needs particular care.
Governance programs often fail by starting too big. The ones that work usually start with the handful of metrics leadership relies on, govern those well, and expand from there. See KPI drift.