There's a growing conversation about how data should be treated as a product in the world of enterprise data. This idea gained significant traction with the rise of the “Data Mesh” architecture a few years ago, which introduced the concept of decentralized data ownership and product thinking.

However, the definition of a data product is still debated. While interpretations vary, most agree that a data product should:

• Be packaged in a way that adds value to the end-user
• Follow a supply-chain process, transforming raw data into a usable outcome
• Serve customers, such as business stakeholders
• Address real-world use cases, including BI dashboards or ML-driven applications
• Be built around domains, microservices, and APIs

There may be other considerations, but this post won't delve into every nuance.

The Synergy Between Data Modeling & Data Catalogs

To understand their roles, let’s first define data catalogs and data modeling tools:

Data Catalog

• A platform that centralizes the management, organization, and discovery of data assets.
• It indexes and classifies data from various sources.
• Capabilities include data lineage tracking, data governance, and business glossary management.
• Ensures data quality, compliance, and fosters collaboration across teams.
  
  

Data Modeling Tools

• Tools that help design, visualize, and manage data models.
• They define the structure, relationships, and rules of data in databases or systems.
• Capabilities include version control, schema generation, and data mapping.
• They enhance collaboration, consistency, and accuracy in data architecture.
• Support data integration, optimization, and alignment with business objectives.‍

Why Both are Essential for Building Data Products

Data catalogs excel at answering the basic question: “What data do we have?”

They create a dictionary of your data assets, making it easier to manage and understand them.

Data modeling tools help visualize how these assets connect with one another. They allow teams to establish the relationships between data entities and better understand how everything fits together in the larger system.

The Analogy: Like Assembling an IKEA Couch

I often use a simple analogy to explain how these tools work together in building data products:

Imagine you’ve bought a couch from IKEA. It could be any product, but for simplicity, let's say it’s a kitchen table.

You unbox the set and find the manual. This manual lists all the necessary parts—screws, bolts, boards, etc.—required to assemble the table.

Although the written instructions are helpful, it might still be difficult to understand how to put everything together. IKEA solves this by including visual images that guide you through the assembly process.

In this analogy:

• The data catalog is like the list of parts in the box.
• A data model is the visual manual that guides you through the assembly.

In Ellie.ai, you can create a visual diagram to show how different data assets (or entities) are related. For example, consider a business concept like a customer. A "customer" isn't just any company—it becomes a customer because of a relationship, such as an invoice or shipment that’s linked to it. The customer is defined by these relationships.

Visualizing Relationships for Better Data Models

This is a simplified example, but it shows how crucial relationships are in data modeling. When you can visualize how business concepts (like customer or product) are interrelated, you're well on your way to creating a solid data product blueprint.

Catalog First, Build Later

In short, the process starts with cataloging your existing data assets. Once that's done, you can build something new, such as a blueprint for creating a BI dashboard or any other data product.

Many of our customers are already using integrations between Collibra and Ellie.ai to achieve exactly this—and they’re seeing great results.