data-governancelisted

# Data Governance for Data Pipelines ## Why Governance Matters Now, Not Later Governance work done retroactively costs 10x what it costs upfront. When an audit arrives or a breach happens, you need to answer three questions fast: *What data do we have? Where does it come from? Who can see it?* If you can't answer these from a system, you'll answer them from a spreadsheet — under pressure, with incomplete information. --- ## PII Classification ### Sensitivity tiers Classify every field that touches personal data before it enters the warehouse. | Tier | Examples | Default access | |---|---|---| | **Public** | Country, product tier, aggregated metrics | All authenticated users | | **Internal** | User ID, subscription status, behavioral events | Analysts and engineers | | **Confidential** | Email, full name, phone number | Restricted to specific roles | | **Restricted** | Payment card data, government ID, health data | Named individuals only, logged | ### Tagging in dbt ```yaml # models/staging/schema.yml models: - name: stg_users columns: - name: user_id meta: pii_tier: internal - name: email meta: pii_tier: confidential pii_type: email_address regulation: [GDPR, PDPA] - name: full_name meta: pii_tier: confidential pii_type: name ``` ### Tagging in BigQuery (policy tags) ```sql -- Assign a policy tag to restrict column-level access ALTER TABLE `project.dataset.u