If your Power BI team manages more than a handful of semantic models, the way those models are stored, versioned, and deployed will directly affect how fast — and how safely — your analytics function can move. Tabular Model Definition Language (TMDL) is Microsoft's answer to a long-standing enterprise pain point: the absence of a human-readable, version-control-friendly format for defining Power BI data models.

As of 2024, TMDL has reached General Availability and is now the default semantic model format in Power BI Desktop. This is not a developer-level footnote. It is a governance shift that affects how your BI team builds, reviews, and ships changes to production — and it deserves executive attention.

What Is Tabular Model Definition Language (TMDL)?

Tabular Model Definition Language (TMDL) is a structured, human-readable format used to define the schema, metadata, relationships, measures, and data processing logic of a tabular data model. Tabular models power the analytical layer behind Power BI, SQL Server Analysis Services (SSAS), and Azure Analysis Services — making TMDL relevant across Microsoft's entire data platform.

Where previous formats serialised the entire model into a single large JSON or XML blob, TMDL breaks the model into a logical folder structure. Each table, perspective, culture, and role lives in its own file. The result is a model definition that teams can read, diff, and merge in a way that was never previously achievable with TMSL (Tabular Model Scripting Language) alone.

TMDL vs TMSL — Why the Format Shift Matters

To understand why Tabular Model Definition Language TMDL is a meaningful step forward, it helps to know exactly what it replaces. TMSL serialises the entire semantic model into a single JSON file. For small models this is manageable. For enterprise-scale models with dozens of tables, hundreds of measures, and complex row-level security, it becomes a governance liability.

The core problem with TMSL at scale

When two developers independently modify different tables in a TMSL file, Git flags a conflict on the same file. Resolving it requires manually inspecting potentially thousands of lines of JSON to find where the changes diverged. This discourages concurrent development and pushes teams toward single-developer ownership of critical model assets — a pattern that creates bottlenecks, increases release risk, and makes auditing changes nearly impossible.

How TMDL resolves this

Because TMDL stores each model object in a separate file, changes to two different tables produce changes to two different files. Git diffs are clean, targeted, and reviewable. Merge conflicts, when they occur, are isolated to a single object's definition rather than the entire model. This architecture directly mirrors the practices that engineering teams apply to production code — which is precisely where enterprise BI should be heading.

The TMDL Folder Structure Explained

When a Power BI project is saved in PBIP format with TMDL enabled, the semantic model is written into a \definition folder. Inside that folder, each major model object — tables, relationships, perspectives, roles, and cultures — is stored as a separate file. A model with 20 tables will produce at least 20 individual files, each containing only the definition relevant to that object.

The practical benefit extends beyond version control. A new team member, an auditor, or an external consultant can orient themselves in an unfamiliar model by browsing the file system rather than parsing thousands of lines of JSON. For organisations managing multiple semantic models across business units, this transparency reduces onboarding time and simplifies compliance reviews significantly.

The TMDL \definition folder in a Power BI PBIP project — one file per model object, one source of truth per table

How TMDL Enables Git Integration and CI/CD for Power BI

The strategic significance of Tabular Model Definition Language TMDL extends well beyond file organisation. When combined with Microsoft Fabric's Git integration, TMDL enables a full software development lifecycle for Power BI semantic models — including feature branching, pull request reviews, automated validation, and controlled deployment to production workspaces.

Fabric Git integration now exports semantic model definitions as TMDL rather than TMSL. This means a change to a semantic model in a development workspace can be committed to a Git repository, reviewed by a senior analyst or model architect, and merged only after approval — giving data leaders visibility and control over model changes that simply did not exist before within the Power BI tooling ecosystem.

For organisations using Power BI consulting services to mature their semantic model layer, this represents a concrete governance capability that previously required expensive custom tooling or workarounds. Teams investing in TMDL-based workflows now will be significantly better positioned as Microsoft continues to deepen Fabric's deployment pipeline capabilities.

The Business Case for Moving to TMDL Now

Executives evaluating whether to prioritise the TMDL migration should consider the cost of the status quo. In organisations where semantic models are managed as TMSL, production changes carry elevated risk: there is no reliable way to see exactly what changed, who changed it, when, or whether it was reviewed before deployment. Rollbacks require manual reconstruction. Concurrent development is effectively impossible without coordination overhead that throttles delivery velocity.

The business case for Tabular Model Definition Language TMDL resolves into three measurable outcomes. First, it reduces deployment risk by making model changes auditable and reversible. Second, it increases development throughput by enabling multiple analysts to work on the same model simultaneously without creating conflicts. Third, it improves model quality by enabling peer review before production — a practice standard in software engineering that remains largely absent from most BI teams today.

For data leaders seeking to justify BI governance investment to their board or finance teams, TMDL adoption provides a tooling-level mechanism for reducing the operational risk that uncontrolled semantic model changes introduce. Numlytics' Power BI Governance Platform provides the operational layer that makes these practices scalable across large enterprise estates.

How to Enable TMDL in Power BI Desktop

For teams not yet using the TMDL format, the migration path is straightforward. In Power BI Desktop, navigate to File > Options and settings > Options > Preview features and enable the option to store semantic models using the TMDL format. Once enabled, saving a project as a Power BI Project file (.pbip) automatically produces the \definition folder structure.

Existing PBIP projects stored as TMSL should be opened in the latest version of Power BI Desktop and re-saved with TMDL enabled. The conversion is non-destructive — your existing report logic and model relationships are preserved. Microsoft has signalled that Fabric Git integration will eventually enforce TMDL as its export format, so teams that defer this migration will face a forced transition on Microsoft's timeline rather than their own.

The earlier this conversion happens, the more time your team has to adjust workflows, update any automation scripts that referenced TMSL output, and build the pull-request review practices that make TMDL's governance benefits real rather than theoretical. If your organisation manages a large estate of Power BI semantic models, our certified Power BI consultants can assess your current portfolio and design a prioritised migration plan.

Next Steps for Your BI Team

The transition to Tabular Model Definition Language TMDL is one of the most consequential changes in Power BI's development model since the introduction of the PBIX format. Organisations that treat it as a developer-level concern and defer executive attention will find themselves operating with a maturity gap as Fabric's deployment capabilities continue to advance. Those that invest now — building version control practices, defining review standards for model changes, and upskilling their analytics teams — will operate with a structural governance advantage that directly reduces release risk and increases analytical output.

Numlytics specialises in enterprise Power BI transformation — from semantic model architecture and governance to deployment pipeline design and capacity management. We have helped data organisations across the US, UK, Australia, and UAE build the BI foundations that make analytics estates manageable at scale. Speak with a certified Power BI consultant to understand how TMDL fits into your organisation's current model management maturity — and what a practical, low-disruption migration plan looks like.