Data Governance is a formal framework: roles (Data Owner, Steward), policies, metric definitions, access control and quality processes - not a one-off spreadsheet tidy-up. Excel is often a symptom of missing governance (shadow reporting), not the solution. Start with the maturity checklist: /en/data-governance-checklist/.

Diagnosis and quick wins (system map, KPI glossary, v1.0 policies) usually take several weeks depending on system count and business involvement. A full programme with MDM and enterprise catalogue is multi-month. Pricing after AS-IS audit - we discuss ranges in consultation: /en/services/data-governance/.

GDPR does not use the words “Data Governance”, but requires accountability, minimisation, security and processing documentation. DG helps connect that with operational and analytics data (classification, retention, lineage). DPO and Data Owner should collaborate - they are not the same role.

First quick wins (5-10 shared KPI definitions, source map, data owners, basic quality rules) often appear in 4-12 weeks. Full maturity is continuous improvement, not a project with a closing date.

Usually different definitions (session vs order), time zones, cookie consent, processing delay and no shared transaction_id across systems. Analytics DG starts with a business KPI glossary and report lineage - guide: /en/data-governance-guide/#analityka-marketingowa.

A business glossary explains business terms (e.g. “active customer”, “conversion”) and maps them to fields in systems and reports. It does not replace GA4/GTM implementation but gives analysts and marketing the same definition. It complements the Measurement Plan and tag documentation.

When you export raw GA4 or other sources to BigQuery and build custom models (SQL, dbt, Looker/PBI). Then governance covers table naming, quality tests, retention and IAM permissions - without that, the warehouse quickly becomes a data swamp.

ML and LLM in the company need trusted, described and current data (garbage in, garbage out). DG defines training sources, PII classification, DPIA and responsibility for quality. That is a prerequisite for sensible AI in reporting and automation.

No. Many organisations start with Notion/Confluence + RACI sheet + quality rules in existing pipelines. Enterprise catalogues (Collibra, Atlan, Alation, DataHub, OpenMetadata) make sense with many teams and systems - after processes stabilise, not on day one.

A catalogue describes technical assets (tables, columns, pipelines, technical owners). A glossary describes business meaning and maps terms to fields. They complement each other; without a glossary, a catalogue is just a “database inventory”.

A golden record is one approved version of an entity (customer, product) merged from multiple systems. You need it with duplicates in CRM/ERP/e-commerce and conflicting reports. Start with consistent product ID types (EAN, SKU, GTIN) before full MDM.

A Data Owner (usually from the business) is accountable for definitions and priorities in a domain (e.g. product, customer). A Data Steward operationally maintains quality, documentation and escalations. Without an Owner, a Steward has no mandate to change processes.

There is no legal requirement, but a regular forum (e.g. monthly) with business, IT, analytics and DPO resolves KPI definition conflicts and quality fix priorities. Without meeting rhythm, governance usually “fades” after the first project.

Use a maturity model: policies, quality (KPI completeness/accuracy), metadata, MDM, security, culture. The checklist on this site gives a 0-100% score and recommendations per category. Repeat measurement quarterly.

Path: (1) maturity checklist, (2) 5-10 shared marketing KPI definitions, (3) GA4-CRM-shop system map, (4) Consent Mode audit. Resource hub: /en/data-governance-resources/.