Put your Power BI models in Git — and keep shipping

PBIP stores semantic models as TMDL and reports as PBIR: plain text that Git can diff, review and merge. I set up the repository, branching, pull-request checks and deployment so your team gets version control without stopping delivery.

Why teams move from PBIX to PBIP

A PBIX file is a binary. Two people cannot work on it at once, a change cannot be reviewed line by line, and rolling back means finding an old copy on SharePoint. That works for one analyst; it breaks for a team.

Power BI Projects (PBIP) save the same content as folders of text: the semantic model as TMDL and the report as PBIR, which Microsoft now documents as generally available and the default report format. Every measure change becomes a reviewable diff, and Fabric Git integration can sync workspaces with a repository.

The format change is the easy part. The value comes from the workflow around it: a repository layout, branch strategy, pull-request reviews, automated checks for risky DAX and missing descriptions, and a controlled path from development to production.

Signs this applies to you

  • Team members overwrite each other's PBIX changes
  • Nobody can say what changed in a model between two releases
  • Production fixes are made directly in the service
  • Measures have no descriptions and inconsistent naming
  • You want to use AI coding agents on models but have no reviewable source

How the setup sprint works

  1. 1

    Assess

    Run the PBIP readiness checker and review current tooling (Azure DevOps or GitHub, deployment pipelines, workspace layout).

  2. 2

    Structure

    Repository layout, naming conventions, .gitignore and branch strategy agreed with the team.

  3. 3

    Automate

    Pull-request checks for DAX anti-patterns, naming and documentation, plus deployment from main to test and production.

  4. 4

    Migrate and train

    Convert the first models, run a real change through the full workflow together, and document it.

What you get

  • Repository with agreed structure, conventions and branch policy
  • First models converted to PBIP/TMDL (and PBIR reports)
  • Pull-request quality checks running in your CI
  • Deployment path from development to production
  • Short team playbook for day-to-day changes

Fixed-Scope Engineering Sprint — Scoped and priced after a short call. Priced by number of models and the CI platform. Includes a hands-on session with your team. See all pricing.

Related reading

Frequently asked questions

Do we have to stop using PBIX files?
No. PBIX remains the single-file format and many people keep using it for personal work. PBIP is for content a team maintains together and wants under version control.
Azure DevOps or GitHub?
Both work with Fabric Git integration and with pull-request checks. The right choice is usually whichever your organisation already uses for code.
Can we start with one model?
Yes — and you should. The sprint converts a first model end-to-end so the team learns the workflow on real work before scaling it.

Sources: Power BI Desktop project report folder (Microsoft Learn).