Why domain ownership without governance recreates the old problem differently
If every domain defines its own data contracts, quality standards, and access controls independently, consumers face the same integration pain as before — just distributed across more teams instead of concentrated in one data platform team. The mesh promised autonomy with consistency; without governance, it delivers autonomy without consistency.
This is precisely why the original data mesh writing treats federated computational governance as one of its four core principles, not an optional add-on — a detail frequently lost in implementation.
What governance-first mesh adoption actually looks like
Defining a minimal, mandatory global standard — schema format, quality SLAs, access control pattern — before domain teams start publishing data products, then giving domains autonomy within that standard, preserves the benefit while avoiding the fragmentation. Sequencing it the other way around is the recurring failure mode.
Data mesh without federated governance isn't a lighter-weight data platform. It's the old integration problem, now owned by nobody in particular.MNB Data Practice
Key actions
- Define a mandatory minimal global data standard before domain rollout
- Treat federated governance as a core principle, not a later phase
- Give domains autonomy within the standard, not instead of it
- Audit consumer integration pain as an early warning signal, not an afterthought