Every organization needs a data catalogdata catalogA centralized inventory of an organization's data assets, enriched with metadata, that helps people find, understand, and trust the data they need.Voir la définition complète →. Most organizations buy one and watch it gather dust.
The data catalogdata catalogA centralized inventory of an organization's data assets, enriched with metadata, that helps people find, understand, and trust the data they need.Voir la définition complète → failure pattern is predictable: leadership gets excited about the promise of "a single place to find all your data." A tool is selected (Collibra, Alation, DataHub). Implementation begins. Engineering teams document the technical metadata automatically. Business users are invited to write definitions. Six months later: the technical is there, most business is missing, and adoption is at 12%.
The tool isn't the problem. The adoption strategy is the problem.
A data catalogdata catalogA centralized inventory of an organization's data assets, enriched with metadata, that helps people find, understand, and trust the data they need.Voir la définition complète → serves four functions:
1. Discovery: Business users and analysts can search for data assets (tables, dashboards, ML models, APIs) using business terms rather than technical names. "Show me everything related to customer churncustomer churnChurn rate is the percentage of customers or revenue lost over a period. It measures how fast a business loses its existing customer base.Voir la définition complète →", the catalog surfaces relevant tables, dashboards, and documentation.
2. Context: For each asset, the catalog shows: who owns it, how fresh it is, how many people use it, what it's related to (Data lineageData lineageData lineage maps how data moves and transforms across systems, from origin to consumption, showing where it came from, what changed it, and where it goes.Voir la définition complète →), known quality issues, and business definitions of key fields.
3. Governance: The catalog is where data classifications live (PII, confidential, public), where access policies are attached, and where data certification (this dataset has been reviewed and is trustworthy) is managed.
4. Collaboration: Data stewards document definitions. Analysts leave comments and questions. Data owners approve or reject proposed definitions. The catalog becomes a living document, not a static registry.
Collibra: Enterprise-grade, strong governance workflows, built-in data policy management, compliance features. Best for large organizations in regulated industries (banking, insurance, healthcare). Implementation complexity is high; typically requires a dedicated implementation partner. Cost: high six figures annually.
Alation: Strong on active metadatametadataDonnées sur les données, informations décrivant le contexte, la structure, la provenance et les caractéristiques d'un asset de données (auteur, date, format, source, définition)., it tracks how data is actually being used, surfaces popular queries, identifies subject matter experts by usage patterns. Popular in mid-market companies and data-mature tech organizations. Excellent self-service analytics support.
DataHub (LinkedIn, open-source): Free, highly customizable, strong integration with modern data stacks (dbt, Airflow, Spark). Requires engineering investment to deploy and maintain. Growing rapidly in organizations with strong data engineering capability. Cost: engineering time rather than license fees.
Atlan: Built for the modern data stack. Strong integrations with dbt, Looker, Snowflake. Collaborative features (Slack-like discussions on data assets). Fastest-growing in this space. Good for organizations using modern, cloud-native tooling.
Vérification des acquis
1. According to the lesson, what is the typical adoption rate of data catalogs six months after implementation in the failure pattern described?
2. Which data catalog is described as best suited for large organizations in regulated industries like banking and healthcare?
3. The lesson mentions data lineage as part of the 'Context' function. What does data lineage primarily help users understand?
4. Select ALL correct statements about the four functions a data catalog serves according to the lesson.
Sélectionnez toutes les réponses correctes.
5. Select ALL correct statements about DataHub based on the lesson.
Sélectionnez toutes les réponses correctes.
A data catalogdata catalogA centralized inventory of an organization's data assets, enriched with metadata, that helps people find, understand, and trust the data they need.Voir la définition complète → with low adoption is worse than no data catalogdata catalogA centralized inventory of an organization's data assets, enriched with metadata, that helps people find, understand, and trust the data they need.Voir la définition complète →, it creates false confidence that governance exists while the data landscape remains undocumented.
The five adoption failure modes:
1. Documentation burden placed on engineers: Engineers will document technical metadatametadataDonnées sur les données, informations décrivant le contexte, la structure, la provenance et les caractéristiques d'un asset de données (auteur, date, format, source, définition). because tools automate it. They will not spend hours writing business definitions. Don't make them.
2. No integration into existing workflows: If analysts have to open a separate tool to look up data definitions, they won't. The catalog must surface where people already work: in Slack (catalog search bot), in their BIBITechnologies and processes that turn raw data into actionable insights via reporting, dashboards and analysis, so teams can decide based on facts rather than intuition.Voir la définition complète → tool (Looker or Tableau integration), in the Data warehouseData warehouseA central repository that consolidates data from many source systems into a structured, query-optimized store designed for analytics, reporting, and business intelligence.Voir la définition complète → UI.
3. Starting with everything: Trying to document every dataset before launch means you launch with mediocre documentation of thousands of assets instead of excellent documentation of hundreds. Start with your 20 most-used datasets and make them perfect.
4. No curation incentives: Why would a data stewarddata stewardA business-side owner responsible for the quality, consistency and appropriate use of data in their domain.Voir la définition complète → spend time writing catalog definitions? Create recognition, a "catalog contributor of the month" award, integration into performance reviews for data roles, a public quality score for each domain.
5. Wrong success metric: "Number of assets documented" is a vanity metric. Track "percentage of data consumers who found what they were looking for in the catalog", a survey-based metric that actually measures whether the catalog is working.
ING Bank implemented Collibra as their enterprise data catalogenterprise data catalogA centralized inventory of an organization's data assets, enriched with metadata, that helps people find, understand, and trust the data they need.Voir la définition complète → across 40+ countries. Their success factor: they didn't implement a catalog, they implemented a data governancedata governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.Voir la définition complète → operating model that happened to use Collibra as the tool.
They defined data ownership first. They trained Data Stewards second. They built the governance workflows third. The tool last. This sequence, people and process before technology, is consistently the differentiator between catalog implementations that work and those that don't.