Enterprise architecture best practices

Established patterns.
Expert implementation.

We bring proven industry architectures together to implement enterprise data warehouses that support reporting, managed self-service BI, and AI through conversational BI. Our expertise is in selecting, connecting, and implementing these established practices for your business.

The medallion structure organizes the journey from source data to analytics. Within it, we apply automated ingestion in Bronze, Data Vault integration in Silver, and enterprise Kimball dimensional modeling in Gold. A shared semantic layer brings consistent business meaning to the result.

Discuss your architecture Explore the model

From source to shared understanding

Every layer
has a purpose.

Each layer uses an established approach for a different job. This is a combination we implement successfully; medallion architecture itself does not prescribe Data Vault or Kimball. The patterns and tools are selected to fit your requirements.

  1. Sources

    Business systems

    Operational applications, databases, files, and APIs.

  2. Bronze

    Automated ingestion

    Automated source capture with fidelity and traceability.

  3. Silver

    Data Vault integration

    Hubs, links, and satellites; Raw Vault history and Business Vault rules.

  4. Gold

    Enterprise Kimball

    Facts and conformed dimensions aligned across business processes.

  5. Semantic

    Shared semantic views

    Reusable business definitions, measures, relationships, and access rules.

  6. Consumers

    Many ways to use it

    Enterprise reporting, managed self-service BI, and conversational BI.

Across every layer

  • Metadata, catalog & lineage
  • Orchestration & automation
  • Data quality & reconciliation
  • Governance & security
  • Monitoring
  • Version control & CI/CD
Established industry patterns, brought together in one implementation. Bronze, Silver, and Gold form the medallion layers; shared semantics serve their analytical output. Automation supports every layer. WhereScape is an optional tool for implementing these patterns.

These architectures provide complementary foundations for governance: source lineage and Data Vault historization support auditability, Kimball conformed dimensions support consistent analysis, and shared semantic models govern reusable business definitions. Metadata-driven automation helps apply those standards consistently as the warehouse grows.

Bronze / automated ingestion

Preserve what the source provided.

The Bronze layer applies established ingestion and staging patterns to preserve source fidelity. Metadata-driven automation makes acquisition repeatable, providing a consistent foundation as new sources are added.

Separating source capture from business interpretation supports traceability and reprocessing. It gives the integration layer a dependable starting point while keeping routine ingestion efficient.

  • Consistent, repeatable source onboarding.
  • Traceable source data for downstream integration.
  • Less repetitive development through automation.

Silver / Data Vault architecture

Integrate the enterprise and its history.

Data Vault's hubs, links, and satellites separate business keys, relationships, and descriptive history. The Raw Vault preserves source-aligned history and provenance; the Business Vault adds reusable business rules and derivations. Together, they support an auditable integration layer that can evolve as systems change.

Our certified Data Vault practitioners apply established modeling, architecture, and delivery standards to help realize these benefits. Repeatable patterns support automation, while traceable history and reusable rules strengthen governance across the warehouse.

  • Traceable history across changing source systems.
  • Extensible integration without redesigning every downstream report.
  • Reusable rules and repeatable, parallel loading patterns.

Gold / enterprise Kimball architecture

Connect analysis across business processes.

Kimball dimensional modeling organizes business processes into facts at a declared grain and dimensions that describe their context. The enterprise bus matrix maps processes to shared dimensions, guiding incremental delivery within an enterprise design.

Conformed dimensions give teams a consistent way to analyze shared concepts such as customer, product, and date. Gold translates integrated history into understandable analytical structures, with historical dimension behavior chosen to match reporting needs.

  • Business-friendly structures suited to BI queries.
  • Consistent comparisons across processes using conformed dimensions.
  • Incremental delivery of marts that fit the enterprise model.

Semantic / shared business meaning

Govern the definitions. Enable self-service.

Semantic layer architecture publishes shared business views over the analytical data: agreed measures, KPIs, relationships, hierarchies, and business names. Documented, versioned models keep that logic reusable across reports and tools, with access controls supported by the chosen platform.

Managed self-service BI separates ownership of shared models from report creation. Central teams maintain trusted definitions and permissions; business teams create reports and explore the models within that framework.

  • Shared metrics without rebuilding logic in every report.
  • Business flexibility with consistent definitions and governed access.
  • Business context for conversational BI over the same foundation.

Managed self-service analytics

Define once.
Reuse across teams.

Managed self-service BI is an established operating model: shared data and semantic models are governed centrally, while report creation is distributed to business teams. Maintaining definitions, measures, documentation, and access rules together makes those models reusable.

Departments build thin reports that connect to shared models, and analysts explore them through tools such as Excel. Conversational BI adds natural-language access to this foundation through configured data agents.

That separation lets each team shape its view of the business while keeping core definitions consistent.

Explore semantic model delivery

Illustrative implementation / Microsoft stack

Sales semantic model

Agreed measures · relationships · hierarchies · access rules

Power BI

Team reports

Thin reports use shared measures.

Excel

Managed self-service

Explore shared definitions with governed access.

Conversational BI

Ask the data

Ask questions in business terms.

One illustrative domain model supports multiple ways of working. An enterprise can publish a portfolio of reusable models for different business domains.

How the architecture enables AI

Give AI the context
behind the data.

The enterprise warehouse serves reporting and analysis first, and the same foundation also supports conversational BI. Automated capture preserves source context. Data Vault connects business keys, relationships, and history. Kimball models organize that history for analysis.

Semantic models supply governed measures and calculations. Ontologies describe business entities and their relationships. Data agents use that context to interpret questions, query the right sources, and return answers in business terms.

We connect conversational BI to this governed foundation so business questions draw on shared definitions and relevant context. Evaluating answers against agreed measures helps teams use the capability with confidence.

Explore AI over your business data

A reusable data foundation

Connected data. Preserved history.

  • Bronze / automated capture
  • Silver / Data Vault
  • Gold / enterprise Kimball

Measures & calculations

Semantic models

Define sales revenue, order counts, and reporting periods.

Entities & relationships

Optional ontologies

Describe how customers, orders, products, and categories connect.

Illustrative conversational BI experience

Data agents

Use configured sources and business context to answer questions within the user's permitted data access.

How has sales revenue changed by product category and quarter?

Question Select context Query Answer

Illustrative sales context. Agents query shared measures through semantic models; ontologies add entity and relationship context where useful.

Designed for delivery

Govern the model.
Give reporting room to move.

Separate model and report lifecycles, with development, staging, and production environments. Promote changes with validation, clear dependencies, and repeatable deployment.

Scroll across to compare delivery environments.

Separate model and report delivery across three environments
Delivery trackDevelopmentStagingProduction
Shared modelsBuild & versionValidate & reviewPublish governed products
Departmental reportsCreate & iterateCheck model connectionsRelease business views

Reports connect to the appropriate shared model in each environment. Model definitions and report layouts can evolve through their own release processes.

Automate the standards

Choose tools that fit your team.

Automation is a delivery practice across the architecture. We have considerable experience with WhereScape and custom templates, and are a certified WhereScape partner. WhereScape is optional: we can implement the same architecture using your platform's pipelines, versioned SQL, and other automation tools.

Explore our WhereScape experience

Adapt to your platform

Connect design to the platform.

We can implement these patterns on your chosen data platform, adapting storage, integration, and deployment to fit. Azure Synapse and Microsoft Fabric are particular areas of depth for our team.

Evaluate the AI experience

Test answers against the model.

Use representative business questions to check agent responses against shared measures and definitions. Review source access and refine the configuration as models and business needs evolve.

Industry patterns into practice

Build the foundation
your next use case can reuse.

We can help assess your platform and bring these proven approaches together for enterprise BI, managed self-service, and conversational BI.

Let's talk architecture