Integrability

Definition

Integrability is the ability of a data model to connect, coexist, and interact with other models, sources, systems, and analytical domains without losing semantic coherence or logical stability.

An integrable model can be coupled to new environments,
but it does not become contaminated.

In Power BI, this property ensures that the model can integrate with:

• new data sources,
• other semantic models,
• external layers (Excel, APIs, Dataflows, Fabric),
• or broader analytical ecosystems,

without breaking its internal meaning.

Nature

Structural, semantic, and architectural.

Integrability does not depend on a single table or measure,
but on the overall design of the model and its ability to coexist with other systems.

It emerges when:

• keys are well defined,
• dimensions are reusable,
• measures do not rely on local assumptions,
• and the model respects clear standards.

Function

To allow the model to stop being an isolated silo and become a reusable component of the data ecosystem.

In practice, integrability enables:

• combining models without massive rewrites,
• incorporating new sources without redesigning everything,
• scaling analysis from local to organizational levels.

Consequence

• Greater model reuse.
• Less friction when scaling solutions.
• Reduced semantic duplication.
• Better alignment across teams and domains.

An integrable model connects and adds value;
a non-integrable model isolates and fragments.

Signals of Integrability

• Uses shared dimensions (Date, Product, Customer).
• Keys are consistent across systems.
• Measures continue to work as the model expands.
• Can connect to other models without redefining concepts.
• Integrates with Dataflows, Fabric, or composite models without breaking logic.

Signals of Non-Integrability

• Relies on local keys with no external correspondence.
• Uses different meanings for the same concept.
• Measures assume “this model and nothing else”.
• Integrating another source requires rebuilding the model.
• Each report creates its own version of the truth.

This is not autonomy; it is structural isolation.

Conceptual Example in Power BI

Non-integrable model (❌)

• DimDate created differently in each report.
• Products with different IDs depending on the source.
• Measures dependent on local-specific columns.

Integrable model (✅)

• Conformed dimensions.
• Decoupled facts.
• Context-based measures, not origin-based.

This measure:

• works in composite models,
• remains valid when integrating new sources,
• does not depend on a closed structure.

Integrability and Composite Models

Integrability is a prerequisite for:

• composite models,
• DirectQuery + Import,
• shared semantic models,
• Fabric / OneLake scenarios.

Without integrability, these approaches collapse.

Interactions

• Consistency → ensures stable meaning.
• Equivalence → enables coherent results across paths.
• Semantic Elasticity → adapts meaning to context.
• Integrability → allows models to coexist.

Without integrability, the other properties remain isolated.

Samples

Sample 1 — Conformed dimension (shared Date)

Problem (non-integrable ❌)
Each model has its own DimDate.

• DimDate_Sales
• DimDate_Budget
• DimDate_Operations

👉 Integrating models = immediate conflict.

Integrable solution (✅)
A single conformed DimDate:

• DateKey
• Date
• Year
• Month
• MonthYear

📌 Result:
Models can be joined, compared, and scaled without redefining time.

Sample 2 — Measures decoupled from origin

Non-integrable (❌)

If FactSales_CRM is later integrated, the measure fails.

Integrable (✅)

📌 Result:
The measure works even if the origin changes or expands.

Sample 3 — Multiple fact integration

📌 Result:
Both facts share conformed dimensions → natural integration.

Sample 4 — Composite model (Import + DirectQuery)

Total Sales Unified :=

📌 Result:
The user does not know or care where the data comes from.
The meaning remains intact.

Sample 5 — Integration between semantic models

Model A (Finance)
Model B (Operations)

📌 Result:

• A financial KPI can be crossed with operational metrics.
• Without duplicating logic.
• Without redefining concepts.

Sample 6 — Integrability vs silos (antipattern)

Antipattern (❌)

• its own DimProduct,
• its own definition of “Sales”,
• its own calendar.

👉 Integrating = rebuild everything.

Integrable pattern (✅)

• Conformed dimensions
• Reusable measures
• Shared semantics

👉 Integrating = connect, not rebuild.

Synthesis

An isolated model may function.
An integrable model can scale.