Ravindra BagaleCourses & study guides मराठी Track your progress

2. Installation and Interface Tour

2.7 PBIX, PBIP and the TMDL Project Folders

By Last updated:

Start with a problem

Shraddha reviews Ravindra's model changes, but with a .pbix she cannot see which measure changed.

Explain simply

A Power BI Project (.pbip) saves the report and the semantic model as text folders. The model is stored as TMDL text, so Git can show a line-by-line diff.

Show the data

Measure Before After
AOV DIVIDE([Total Sales], COUNTROWS(Orders)) DIVIDE([Total Sales], [Delivered Orders])
Result on practice data ₹262.50 (2,100 ÷ 8) ₹300.00 (2,100 ÷ 7)

Fictional sample data. Practice file: visuals_orders_apr_may_2026.csv (source: src/practice/).

Demonstrate

  1. Save the normal file as MHQuickCommerce.pbix (single-file copy).

  2. File › Save as › Power BI project files (*.pbip) › an empty folder.

    Your result should look like this:

    Folder
    MHQuickCommerce.pbip, MHQuickCommerce.Report/, MHQuickCommerce.SemanticModel/

  3. Close Desktop › open the SemanticModel folder in a text editor › search AOV.

    Your result should look like this:

    Text found
    measure AOV = DIVIDE([Total Sales], COUNTROWS(Orders))

  4. Change the measure to use [Delivered Orders] in Desktop › save › run git diff.

    Your result should look like this:

    Diff
    − COUNTROWS(Orders) + [Delivered Orders] (one line)

  5. Commit the .pbip, .Report and .SemanticModel folders; ignore cache and local settings with .gitignore.

Show the result

Your result should look like this (fictional sample):

Check Value
Lines changed in the diff 1 (the AOV expression)
AOV before ₹262.50
AOV after ₹300.00

Explain what happened

Because the model is text, Git showed exactly the changed expression. AOV moved from ₹262.50 (cancelled order counted) to ₹300.00 (delivered orders only).

Explore further

  • Menu labels for PBIP and PBIR move between Desktop versions; the idea stays the same.

Detailed reference for this lesson:

A .pbix is one file. When you work with a team in a job, you need folders instead. This is version-dependent: Microsoft keeps changing the project format in Desktop, so menu names may look slightly different on your machine.

A .pbix file is one package. It holds the Power Query steps, the semantic model, imported data and the report. It is the right file while you are learning and for a report that one person owns. It is a poor file for Git, because most of it is binary, so a small change looks like the whole file changed.

A Power BI Project (.pbip) saves the same work as text folders. File › Save as › Power BI project files (*.pbip) (the exact label moves between Desktop versions). You get a folder something like this:

Piece What it is
Something.pbip A small text pointer that opens the project in Desktop
Something.Report/ The report definition (pages, visuals). Newer files use the PBIR layout; older ones use report.json
Something.SemanticModel/ The model. definition.pbism points at the model, and the tables, relationships and measures are TMDL (Tabular Model Definition Language) text, often under a definition/ folder
.gitignore You should ignore cache and local settings so they are not committed

TMDL view (the icon next to Model view in current Desktop) shows that same model as script. You can read a measure or a relationship there. Bulk edits in TMDL are for people who already understand the model; a wrong edit can stop the file from opening. Prefer the ribbon for everyday work, and use TMDL when you need to review a change in source control.

Steps – save the fictional quick-commerce file as a project

  1. Save your normal file as MHQuickCommerce.pbix first, so you still have a single-file copy.
  2. File › Save as and choose the Power BI project type. Pick an empty folder.
  3. Close Desktop and open the folder in a text editor. Find Total Sales inside the semantic model folder. You should see the DAX as text, not inside a binary file.
  4. If your team uses Git, commit the .pbip, the .Report folder and the .SemanticModel folder. Do not commit credentials.

Worked example. Ravindra Bagale keeps the Pune report, and Shraddha Bagale reviews the Nashik measures. With a .pbix, Shraddha cannot see which measure changed. With PBIP, the Git diff shows only the Gross Margin % measure text. The data is still the fictional Maharashtra sample (Pune, Mumbai, Nashik, Sambhaji Nagar, Kolhapur, Nagpur). PBIP does not remove imported data from your machine; follow your company rule before you push a folder that contains a local cache.

Ravindra Bagale's Tip

Friends, many students turn on PBIP and then e-mail the whole project folder, including a local cache, to a classmate. A project is for source control, not for sending data. Share a published app, or a .pbit when the other person should reload the data themselves. Pay attention!

Fix common mistakes

Wrong result Reason Correction
Whole folder e-mailed with a local cache PBIP treated as a sharing format Use Git; share apps or .pbit
Credentials committed Local settings not ignored Use .gitignore; never commit secrets
TMDL edit breaks the model Bulk edit without understanding Edit in Desktop or test in a copy

Practise independently

Practice task

With the practice data, what would AOV be if it divided by distinct cities instead (a wrong formula)? Why would a reviewer catch it in the diff?

Hints
  • 3 cities.
Solution

₹2,100 ÷ 3 = ₹700. The diff shows the changed DIVIDE line, so the reviewer sees the wrong denominator immediately.

Think like a professional

Use PBIP when more than one person changes a model. Review measure changes like code.

Thodkyaat sangaycha tar (quick recap)

Visuals in Report view, data checks in Table view, and relationships in Model view. If a pane closes, bring it back from the View tab. Now let us move on – it is time to bring in data!

Ravi Sir's Cards for this lesson

PBIP project and TMDL folders. A .pbix is one closed file. Git cannot show what changed inside it, so team work is hard.Revision card

मराठी cards