Once you've profiled your data, the next crucial step is transformation. Power BI empowers you to clean, reshape, and structure data with the help of its intuitive Power Query Editor. Whether you're removing unnecessary columns or pivoting your entire dataset, these transformations make your data analytics-ready — without ever touching the original source.

"Power Query transformations are non-destructive — your original data remains untouched while you shape a working version tailored to your needs."

🔧 Common Data Transformation Actions

Power Query Editor exposes a rich set of transformation operations through its ribbon — no code required for most tasks. Here are the most commonly used ones:

🗑️
Remove Columns / Rows
Strip out irrelevant, empty, or duplicate data that would clutter your model.
✏️
Rename Columns
Give columns meaningful names — "Amt" becomes "Sales Amount", "Dt" becomes "Order Date".
🔢
Change Data Types
Ensure dates are typed as Date, numbers as Decimal/Integer, and text as Text. Wrong types break DAX calculations.
🔽
Filter Rows
Keep only the records that matter — filter by date range, region, product category, or any condition.
✂️
Split Columns
Break "Full Name" into "First Name" and "Last Name", or split "Product - Size" into two usable columns.
🔗
Merge Columns
Combine "City" and "State" into a single "Location" column for cleaner visualisation.
🔄
Pivot / Unpivot Columns
Reshape flat wide tables into tall, analysis-friendly formats — or vice versa. Essential for poorly structured source data.
📦
Group By
Aggregate rows — sum sales by region, count orders by product, or average quantity by category.

⚙️ Applied Steps — The Magic Recorder

Every transformation you perform is recorded in the Applied Steps pane on the right side of Power Query Editor. Think of it as your personal data-cleaning history — a transparent audit trail of every operation applied to your data.

  • Rename any step to make the pipeline self-documenting
  • Double-click a step to review or edit the transformation settings
  • Delete a step to undo it — the remaining steps re-run automatically
  • Reorder steps when needed (with caution — some steps depend on earlier ones)
  • Each step generates an M language formula — visible in the formula bar
💡 Key Principle Applied Steps are executed top to bottom in sequence. If you rename a column in Step 3, any step after that which references the old column name will break. Always be mindful of step order when editing existing pipelines.

🔄 Real-World Example — Sales Report Cleanup

Imagine you're working on a report for a retail clothing company. You receive a CSV file with thousands of rows from multiple store branches. Here's a typical transformation pipeline:

1
Remove irrelevant columns
Delete "Internal Notes" and "ProcessedBy" — not needed for sales analysis.
↓
2
Fix data types
"Sale Date" is stored as text — change it to Date type so time intelligence DAX works correctly.
↓
3
Filter rows
Keep only data from the last 2 years — filter on Sale Date ≥ today minus 730 days.
↓
4
Split column
"Product Name - Size" splits into two columns: "Product Name" and "Size" — by delimiter.
↓
5
Merge columns
Combine "City" and "State" into a single "Location" column separated by ", " for cleaner map visuals.
↓
✓
Close & Apply
Data is clean, consistent, and loaded into the model — ready for relationships and DAX measures.

🧠 Analogy — Data as Lego Blocks

Think of each row of data as a Lego piece. Raw data might have broken, mismatched, or oddly shaped pieces. Transformation is like fixing the pieces — cutting some, merging others, snapping them into the right orientation — so you can build a stable, accurate model.

Without proper preparation, your Lego castle will collapse under its own weight. 🏰 The same is true for Power BI reports built on messy, untyped, or duplicated data.

📌 Summary — Transformation Principles to Remember

  • Non-destructive by design — Power Query never modifies your original data source. All transformations create a computed view.
  • Applied Steps are sequential — steps execute top to bottom. Edit with awareness of downstream dependencies.
  • Data types matter — wrong types cause DAX errors and incorrect aggregations. Always verify types after loading data.
  • Transform before you model — clean data in Power Query, not in DAX. It is more efficient and easier to maintain.
  • Transformation = trust — reports built on clean, well-shaped data earn credibility with stakeholders.

🧠 Try It Yourself

Open the Sales dataset in Power Query and complete these tasks:

  1. Filter rows based on a condition
    Keep only orders where the Sales Amount is greater than 1000. Check the Applied Steps pane — notice the new "Filtered Rows" step.
  2. Split a column
    If your dataset has a column like "Product - Size", split it by the delimiter " - " into two separate columns. Rename both clearly.
  3. Rename an Applied Step
    Right-click any step in the Applied Steps pane and rename it to something descriptive like "Remove Internal Columns" instead of the default "Removed Columns".
Up Next in Chapter 4 4.4 Data Modeling — Fact and Dimension tables, star schema vs snowflake schema, relationship cardinality, and why a clean model is the foundation of every great Power BI report.