Data mesh is an organisational and architectural approach for large companies where a single central data team becomes a bottleneck.
The Four Principles
- Domain ownership: the teams closest to the data — sales, logistics, payments — own and publish their data, rather than a central team.
- Data as a product: published data is treated as a product with users: documented, discoverable, reliable, with a named owner and quality commitments.
- Self-service data platform: a central platform team provides shared tools and infrastructure so domain teams can build and publish data products without being infrastructure experts.
- Federated computational governance: common standards — interoperability, security, privacy, definitions — agreed centrally and enforced automatically where possible.
Why Organisations Adopt It
- Central teams can't keep up with demand in large organisations.
- Domain experts understand their data best.
- Clear ownership improves quality.
Challenges
- Requires data skills within domain teams.
- Risk of inconsistency without strong standards.
- Significant organisational change, not just new technology.
- Can be overkill for smaller organisations.
Is It Right for You?
Data mesh suits large organisations with many distinct domains and a mature data culture. Smaller organisations usually benefit more from a well-run central platform — while still borrowing the idea of treating data as a product with clear owners.
Start With the Product Mindset
Whatever the architecture, documented, owned, quality-checked data products make data easier to trust and use.