Treating data as a product in the era of GenAI

Data mesh architecture

Many organizations are adopting data mesh, a federated data architecture that allows them to meet the growing demand of data across the enterprise. Data mesh architecture is based on four key principles:

1. Domain-oriented data
2. Data as a product
3. Self-service infrastructure
4. Federated computational governance

Our report explores the importance and power of the second principle, treatment of data as a product. In the data mesh, data isn’t just an asset; it’s a product with well-defined owners, consumers, and quality standards. Treating data as a product motivates domain teams to manage their data as a product and treat the rest of the organization as their customer.

What does it mean to treat data as a product?

As organizations adopt data mesh, a transformative paradigm for their data architecture, the treatment of data as a product is at its heart. Treating data as a product calls for breaking down the data monolith into smaller data products, inspired by the “API-first” modular, reusable design principles adopted for applications over the past decade. This evolution will lead to data being treated as an enterprise asset and managed by source- or consumption-oriented domains.

What is a data product?

Data products are assets that help organizations take control of their data and generate business value. They are shareable and can help unlock the potential of data in a way that benefits internal and external customers.

Organizations can lay the foundation for value delivery by instilling product thinking within data ecosystems. Data products allow them to unlock the power of their vast datasets by empowering analysts to focus on business insights. Here are five ways data products can help an organization:

1. Cater to multiple use cases with standardized key datasets to ensure consistency of the insights across the organization.
2. Increase confidence in data with consistent definitions, usage, and results across teams.
3. Make the standardized and reusable datasets easily searchable in a data marketplace where analysts can share their knowledge and findings.
4. Let analysts focus on producing valuable insights instead of constantly battling with data.
   
5. Reduce IT overhead with a data-simplified architecture. Federate IT development costs back to LOB analysts with a chargeback model.