Data model best practices in Power BI - Step-by-Step Guide

Introduction

A good data model helps your reports run faster and be easier to understand. It organizes your data so you can find answers quickly without confusion or errors.

When you want your sales report to update quickly even with lots of data

When you need to combine data from different sources like Excel and databases

When you want to avoid mistakes caused by duplicate or missing data

When you want to create clear relationships between tables for easy analysis

When you want to make your report easy for others to use and maintain

Steps

Step 1: Open your Power BI Desktop file

- Power BI Desktop main window

You see your current data model and tables

💡 Start with a clean model before adding new data

Step 2: Click on the Model view icon

- Left vertical pane

You see all tables and their relationships visually

💡 Use this view to check how tables connect

Step 3: Select a table and review its columns

- Model view canvas

You see all columns and their data types

💡 Remove any columns you don’t need to keep the model simple

Step 4: Create relationships by dragging a column from one table to another

- Model view canvas

A line appears showing the relationship between tables

💡 Make sure relationships are one-to-many and use unique keys

Step 5: Set the correct data types for each column

- Model view > Column properties pane

Columns have proper types like text, number, or date

💡 Correct data types improve performance and accuracy

Step 6: Hide unnecessary columns from report view

- Model view > Right-click column > Hide in report view

These columns won’t appear in the fields list for report building

💡 Hiding columns reduces clutter for report creators

Step 7: Use star schema design by having fact tables connected to dimension tables

- Model view

Your model is easier to understand and faster to query

💡 Avoid circular relationships and many-to-many where possible

Before vs After

Before

Model has many columns including unused ones, unclear relationships, and mixed data types causing slow report refresh and confusing fields list

After

Model has only needed columns, clear one-to-many relationships, correct data types, and hidden technical columns resulting in faster reports and easier report building

Settings Reference

Data type

📍 Model view > Column properties pane

Defines how Power BI treats the data for calculations and filtering

Default: Text

Relationship cardinality

📍 Model view > Relationship line > Properties pane

Defines how tables relate to each other for accurate data filtering

Default: One to many

Cross filter direction

📍 Model view > Relationship line > Properties pane

Controls how filters flow between related tables

Default: Single

Hide in report view

📍 Model view > Right-click column or table

Controls whether columns or tables appear in the report fields list

Default: Visible

Common Mistakes

Keeping all columns from source tables even if not used

Extra columns slow down the model and clutter the fields list

Remove or hide columns that are not needed for analysis

Creating many-to-many relationships without a bridge table

This can cause incorrect results and slow performance

Use a bridge or dimension table to create one-to-many relationships

Using text data type for numeric or date columns

Power BI cannot perform calculations or time intelligence correctly

Set correct data types like number or date for those columns

Summary

A clean data model improves report speed and clarity

Use one-to-many relationships and star schema design

Remove or hide unused columns and set correct data types

Practice

1. What is the main reason to use a star schema in Power BI data modeling?

easy

A. It hides data from users to improve security.

B. It allows storing all data in one big table.

C. It automatically creates relationships without keys.

D. It simplifies analysis by separating facts and dimensions clearly.

Solution

1. Step 1: Understand star schema purpose

   A star schema organizes data into fact tables (measures) and dimension tables (descriptions) for clarity.

2. Step 2: Recognize benefits in Power BI

   This separation makes analysis easier and faster by simplifying relationships and queries.

3. Final Answer:

   It simplifies analysis by separating facts and dimensions clearly. -> Option D

4. Quick Check:

   Star schema = clear fact and dimension tables [OK]

Hint: Star schema means clear fact and dimension tables [OK]

Common Mistakes:

• Thinking all data should be in one table
• Believing relationships form without keys
• Confusing star schema with security features

2. Which of the following is the correct way to create a one-to-many relationship in Power BI?

easy

A. Link the primary key in the dimension table to the foreign key in the fact table.

B. Link the foreign key in the dimension table to the primary key in the fact table.

C. Create a many-to-many relationship between two fact tables.

D. Use a calculated column to join unrelated tables.

Solution

1. Step 1: Identify keys in tables

   The dimension table has a primary key; the fact table has a foreign key referencing it.

2. Step 2: Create relationship direction

   One-to-many means one dimension row links to many fact rows, so connect primary key (dimension) to foreign key (fact).

3. Final Answer:

   Link the primary key in the dimension table to the foreign key in the fact table. -> Option A

4. Quick Check:

   One-to-many = primary key to foreign key [OK]

Hint: Primary key (dimension) links to foreign key (fact) [OK]

Common Mistakes:

• Reversing keys in relationship
• Using many-to-many where one-to-many is needed
• Trying to join unrelated tables with calculated columns

3. Given a data model with a Sales fact table and a Product dimension table linked by ProductID, what happens if you add a relationship from Sales[ProductID] to Product[ProductID] but the Product table has duplicate ProductIDs?

medium

A. Power BI will create a one-to-many relationship and filter correctly.

B. Power BI will show an error and not allow the relationship.

C. Power BI will create a many-to-many relationship automatically.

D. Power BI will ignore duplicates and create the relationship anyway.

Solution

1. Step 1: Understand relationship requirements

   One-to-many relationships require the 'one' side (Product) to have unique keys.

2. Step 2: Check effect of duplicates

   Duplicates in Product[ProductID] break uniqueness, so Power BI blocks the relationship creation.

3. Final Answer:

   Power BI will show an error and not allow the relationship. -> Option B

4. Quick Check:

   Unique keys required on 'one' side [OK]

Hint: One side keys must be unique to create relationship [OK]

Common Mistakes:

• Assuming duplicates are ignored
• Expecting automatic many-to-many creation
• Thinking relationship will still filter correctly

4. You created a relationship between Customer[CustomerID] and Sales[CustomerID], but your report shows incorrect totals. Which of the following is the most likely cause?

medium

A. The Sales table is missing CustomerID values.

B. CustomerID has duplicates in the Customer table.

C. The relationship is inactive and needs to be activated.

D. The relationship is many-to-many instead of one-to-many.

Solution

1. Step 1: Check relationship status

   Inactive relationships do not filter data automatically, causing wrong totals.

2. Step 2: Confirm activation needed

   Activating the relationship or using DAX USERELATIONSHIP fixes filtering issues.

3. Final Answer:

   The relationship is inactive and needs to be activated. -> Option C

4. Quick Check:

   Inactive relationship causes wrong totals [OK]

Hint: Inactive relationships block filtering; activate them [OK]

Common Mistakes:

• Ignoring inactive relationship status
• Assuming duplicates always cause wrong totals
• Thinking missing values break relationships

5. You have a complex data model with multiple fact tables and dimension tables. To improve performance and clarity, which best practice should you apply?

hard

A. Use a star schema with clear one-to-many relationships and avoid circular dependencies.

B. Combine all tables into one large table to reduce relationships.

C. Create many-to-many relationships between all tables for flexibility.

D. Use calculated columns to join tables instead of relationships.

Solution

1. Step 1: Identify best modeling approach

   Star schema organizes data clearly with fact and dimension tables linked by one-to-many relationships.

2. Step 2: Avoid common pitfalls

   Combining tables or many-to-many relationships can cause confusion and slow performance; calculated columns for joins are inefficient.

3. Final Answer:

   Use a star schema with clear one-to-many relationships and avoid circular dependencies. -> Option A

4. Quick Check:

   Star schema + one-to-many + no circular dependencies [OK]

Hint: Star schema and one-to-many improve clarity and speed [OK]

Common Mistakes:

• Merging all tables into one
• Using many-to-many everywhere
• Relying on calculated columns for joins