Back to Glossary

Data Architecture

How to design a data mesh for a bank

Design a data mesh for banks by implementing decentralized domain-driven data products with embedded governance, enabling self-service analytics while maintaining regulatory compliance through federated ownership across business domains like lending, deposits, and treasury.

Why It Matters

Data mesh reduces time-to-insight by 60-80% compared to traditional data warehouses while cutting data integration costs by 40-50%. Banks achieve 3-5× faster regulatory reporting cycles and improve data quality scores from 70% to 95% through domain ownership. The architecture eliminates single points of failure that cost tier-1 banks $2-4M annually in downtime during regulatory submissions.

How It Works in Practice

  1. 1Identify domain boundaries aligned with business functions like retail banking, corporate lending, and wealth management
  2. 2Establish domain data product owners with P&L accountability for data quality and consumer satisfaction metrics
  3. 3Implement self-serve data platform capabilities including automated data pipelines, discovery catalogs, and standardized APIs
  4. 4Deploy federated computational governance enforcing data classification, retention policies, and access controls through code
  5. 5Create cross-domain data contracts specifying schema evolution, SLA commitments, and backward compatibility requirements
  6. 6Monitor data product health through consumer satisfaction scores, usage analytics, and incident response times

Common Pitfalls

Underestimating regulatory data lineage requirements across domains can trigger BCBS 239 compliance failures during supervisory reviews

Creating domains that mirror existing IT silos rather than true business capabilities leads to continued data integration bottlenecks

Insufficient investment in self-serve platform capabilities forces domain teams to build custom infrastructure, increasing operational complexity by 200-300%

Key Metrics

MetricTargetFormula
Domain Data Quality Score>95%Percentage of domain datasets passing automated quality checks divided by total datasets
Consumer Satisfaction>4.5/5Average rating from downstream consumers of domain data products surveyed quarterly

Related Terms