Power BI Unit Tests

Run DAX unit tests against a locally open Power BI Desktop model — no Power BI Service required.

How it works

The tool connects to the Analysis Services instance that Power BI Desktop runs locally (msmdsrv.exe) via ADOMD.NET. It reads a test-case table from your model, evaluates each measure with its filters, and compares the result against the expected value.

Prerequisites

• Windows (Power BI Desktop is Windows-only)
• Power BI Desktop open with a model loaded
• Python 3.9+
• .NET Framework 4.7.2 (usually already present on Windows)

Installation

From pip

pip install pbi-unit-test

The ADOMD.NET client library is bundled with the package, so no additional setup is required.

Custom ADOMD path (optional)

If you need to use a different version of the ADOMD client library, set the ADOMD_PATH environment variable:

set ADOMD_PATH=C:\path\to\your\adomd\net472

Test table schema

Create a calculated table (or a regular table) in your Power BI model with these columns:

Column	Type	Description
[Measure]	Text	Name of the measure to test
[Filters]	Text	DAX filter context passed to CALCULATE, e.g. Store[Country] = "US"
[WithRounding?]	Boolean	If TRUE, rounds the actual result to 0 decimal places before comparing
[Expected Value]	Decimal	The value the measure should produce

Example calculated table:

DAX

unit_tests_2026_03_22 = DATATABLE(
    "Id", INTEGER,
    "Measure", STRING,
    "Filters", STRING,
    "WithRounding?", BOOLEAN,
    "Expected Value", DOUBLE,
    {
        { 1 ,"Total Sales", "Store[Country] = ""US""", FALSE, 1234567.89 },
        { 2 ,"AOV",         "Store[Country] = ""US""", TRUE,  42 }
    }
)

Power Query

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WMlTSUQrJL0nMUQhOzEktBvKCS/KLUqOd80vzSooqYxVsFWKUQoNjlHQUnIEK8lISi6IjUxOLYm2NDIxMgcrdEnOKU4G0oZGxiamZuZ6FpVKsTrSSEVDI0T+MePMUgDJQI0OKSkEmmhgpxcYCAA==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [Id = _t, Measure = _t, Filters = _t, #"WithRounding?" = _t, #"Expected Value" = _t]),
    #"Changed Type" = Table.TransformColumnTypes(Source,{{"Id", Int64.Type}, {"Measure", type text}, {"Filters", type text}, {"WithRounding?", type logical}, {"Expected Value", type number}})
in
    #"Changed Type"

Usage

Command line

python -m pbi_unit_test <table_pattern> [options]

Arguments

Argument	Description
table_pattern	Table name or glob pattern (e.g. unit_test_*) to run all matching tables
--output-measure [FILE]	Generate a live DAX measure. Omit FILE to print to stdout; provide a path to write to a file
--csv FILE	CSV file to append results to (default: test_results.csv)
--force	Re-run tables already recorded in the CSV, replacing their rows

The process exits with code 0 if all tests pass, 1 if any test fails or no tables matched the pattern — suitable for CI pipelines.

---

Use case 1 — Run a single test table

python -m pbi_unit_test unit_tests

Output:

Connected to: MyReport.pbix

=== unit_tests ===
[PASS ✓] Total Sales | filters: Store[Country] = "US"
[FAIL ✗] AOV | filters: Store[Country] = "US"

1/2 tests passed.

Failed tests:
  AOV: expected 42, got 38.0

---

Use case 2 — Run all tables matching a glob pattern

Useful when you have one test table per sprint or date, e.g. unit_tests_2026_03_*.

python -m pbi_unit_test "unit_tests_*"

Each matching table is run in turn and results are appended to the CSV.

---

Use case 3 — Generate a live DAX measure (print to stdout)

Generates a Test Status measure you can paste directly into Power BI Desktop so you can evaluate tests interactively without re-running Python.

python -m pbi_unit_test unit_tests --output-measure

The generated DAX is printed between two separator lines so it is easy to copy.

---

Use case 4 — Generate a live DAX measure and save to a file

python -m pbi_unit_test unit_tests --output-measure status_measure.dax

The file can be committed to source control alongside your test table.

---

Use case 5 — Track results in a custom CSV file

By default results are appended to test_results.csv in the current directory. Use --csv to choose a different path.

python -m pbi_unit_test unit_tests --csv reports/q1_results.csv

CSV columns: table_name, id, measure, filters, expected, actual, passed, error, dax_query, run_at.

---

Use case 6 — Re-run a table already recorded (--force)

By default the tool skips any table whose name already appears in the CSV, preventing accidental double-runs. Use --force to overwrite the existing rows.

python -m pbi_unit_test unit_tests --force

---

Use case 7 — CI pipeline

The exit code makes it straightforward to fail a build when tests regress.

python -m pbi_unit_test unit_tests || exit 1

Or in a GitHub Actions step:

- name: Run Power BI unit tests
  run: python -m pbi_unit_test unit_tests

---

Use case 8 — Combined: run all tables, generate measure, log to custom CSV

python -m pbi_unit_test "unit_tests_*" --output-measure status_measure.dax --csv results/history.csv --force

Python API

from pbi_unit_test import run_unit_tests, list_dax_measures

summary = run_unit_tests("unit_tests")
summary.print_report()

# Access individual results
for result in summary.results:
    print(result.measure, result.passed)

# Explore measures in the model
print(list_dax_measures())

Project structure

pbi_unit_test/
├── __init__.py        # public API
├── __main__.py        # CLI entry point
├── connection.py      # port detection, ADOMD connection, DAX execution
├── measures.py        # measure listing and expression retrieval
├── runner.py          # test runner, TestResult, TestSummary
└── adom_client/       # bundled ADOMD.NET client library
    └── Microsoft.AnalysisServices.AdomdClient.dll

examples/              # example files
└── status_measure.dax # example status measure DAX

Publishing to PyPI

To publish a new version:

pip install build twine
python -m build
twine upload dist/*