How to Create a Dashboard in Power BI (Step by Step)
· 6 min read
A Power BI dashboard is a single page in the Power BI service made of tiles that you pin from one or more reports. You cannot build one in Power BI Desktop. So creating a dashboard takes three steps: build a report in Power BI Desktop, publish it to the Power BI service, then pin the visuals you want onto a new dashboard. Many people also say "dashboard" for a one-page report full of KPIs. This guide covers both, because the hard part, choosing and checking the numbers, is the same.
Measures for a Brightline Supply Co. sales overview, with the values they return on the sample data:
Completed Revenue =
CALCULATE (
SUMX ( Sales, Sales[Quantity] * Sales[UnitPrice] * (1 - Sales[DiscountPct]) ),
Sales[Status] = "completed"
)
Completed Orders =
CALCULATE ( DISTINCTCOUNT ( Sales[OrderID] ), Sales[Status] = "completed" )
Average Order Value =
DIVIDE ( [Completed Revenue], [Completed Orders] )
Cancelled & Refunded % =
VAR AllRevenue = SUMX ( Sales, Sales[Quantity] * Sales[UnitPrice] * (1 - Sales[DiscountPct]) )
RETURN DIVIDE ( AllRevenue - [Completed Revenue], AllRevenue )| KPI card | Value |
|---|---|
| Completed Revenue | 612.10 |
| Completed Orders | 6 |
| Average Order Value | 102.02 |
| Cancelled & Refunded % | 17.7% |
These four numbers are the top row of the example below. The last one is (744.10 − 612.10) / 744.10: the share of order value that did not turn into revenue.
Dashboard vs report
| Dashboard | Report | |
|---|---|---|
| Where you build it | Power BI service only | Power BI Desktop or the service |
| Pages | One | One or more |
| Data | Tiles from one or more reports and semantic models | One semantic model |
| Slicers and filters | None. You can filter a tile in focus mode, but not save it | Yes, many ways to filter, slice and highlight |
| Drill down | Only if you pin a whole report page as a live tile | Yes |
| Alerts | Yes, on some tile types | No |
| Click a tile | Opens the report it came from | Interacts with the page |
In short: the report is where the analysis lives. The dashboard is a monitoring page that shows the few numbers someone checks every morning and links to the reports for detail.
How to create a dashboard in Power BI
Step 1: Build the report in Power BI Desktop
Load the data, set up the model (a Sales fact table related to Products, Customers and Date) and write explicit measures like the ones above. The model setup is covered in relationships and star schema.
Then build the report page. For Brightline's sales overview:
| Visual | Field | Values on the sample |
|---|---|---|
| Line or column chart | Completed Revenue by 'Date'[Month] | Jan 220.50, Feb 112.10, Mar 279.50 |
| Bar chart | Completed Revenue by Customers[Region] | West 388.10, East 149.00, North 75.00 |
| Bar chart | Completed Revenue by Products[Category] | Cleaning 255.00, Coffee & Kitchen 195.50, Paper & Print 161.60 |
| Four cards | The KPIs above | 612.10, 6, 102.02, 17.7% |
Check the numbers before you design anything. Each chart should add up to 612.10.
Step 2: Publish
Home > Publish in Power BI Desktop, sign in, and choose a workspace. What you can publish and who you can share with depends on your licence and your organisation's settings, so check the official Power BI pages for current details.
Step 3: Pin tiles
In the Power BI service, open the published report. Hover over a visual and select the pin icon. Choose New dashboard, give it a name ("Brightline Sales Overview"), and pin. Repeat for the other visuals and choose Existing dashboard for them. To pin a whole page with its slicers working, use Pin to a dashboard from the report's menu, which creates a live page tile.
Step 4: Arrange and share
Open the dashboard, drag tiles into place, and resize them. Set the tile titles so they make sense without the report next to them. Then share it, or add it to an app for a wider audience.
Dashboard design tips
- Start from the question. "Are we on track this month?" needs revenue against target and a trend, not twelve charts.
- Put the key numbers top left. People read from the top left. Cards for the three or four KPIs go there.
- One message per visual. A bar chart of revenue by region answers one question. If a chart needs a paragraph to explain, split it or drop it.
- Use few colours, on purpose. One neutral colour for everything, one highlight colour for what needs attention.
- Sort bars by value, largest first, unless the categories have a natural order such as months.
- Label units and the business rule. "Completed revenue, 2025" tells the reader cancelled orders are excluded.
- Leave space. A crowded page is harder to scan than one with five well-chosen visuals.
Common mistakes
An AI assistant averages the averages
Ask an AI assistant for an "average order value" card that "works across regions" and you may get:
Average Order Value =
AVERAGEX ( VALUES ( Customers[Region] ), DIVIDE ( [Completed Revenue], [Completed Orders] ) )| Region | Completed Revenue | Completed Orders | Average Order Value |
|---|---|---|---|
| West | 388.10 | 3 | 129.37 |
| East | 149.00 | 2 | 74.50 |
| North | 75.00 | 1 | 75.00 |
| Card (total) | 612.10 | 6 | 92.96 |
The region rows are correct. The card is not: it averages the three regional values, giving North's single order as much weight as West's three. The real average order value is 612.10 / 6 = 102.02. How to spot it: a card should equal the total of its own definition. Divide the revenue card by the orders card by hand, and if the AOV card disagrees, look for AVERAGEX over a grouping column. The fix is the plain DIVIDE measure at the top. The same trap with margins is explained in measure vs calculated column.
A revenue card that counts every status
SUM ( Sales[Line Revenue] ) on a card shows 744.10. It includes a cancelled order (84.00) and a refunded one (48.00). Nobody notices, because the card has no breakdown. Write the business rule into the measure and put it in the card title. DAX CALCULATE explained shows how.
Expecting slicers on a dashboard
Dashboards cannot be filtered or sliced. If users need to slice, give them the report (or pin a live report page). Pinned tiles show the visual with the filters it had when you pinned it.
Too many tiles
A dashboard that tries to show everything gets checked by no one. If it does not fit on one screen without scrolling, cut it.
FAQ
What is the difference between a dashboard and a report in Power BI?
A report has one or more pages built on one semantic model, and you can slice and filter it. A dashboard is one page in the Power BI service with tiles pinned from one or more reports. It cannot be sliced, but it can show tiles from several models and supports alerts.
Can I create a dashboard in Power BI Desktop?
No. Dashboards exist only in the Power BI service. In Desktop you build reports, then publish and pin their visuals to a dashboard.
How do I make a KPI dashboard in Power BI?
Write each KPI as an explicit measure, check it by hand, build a report page with cards for the KPIs and one or two trend and breakdown charts, publish it, and pin those visuals to a dashboard.
How many visuals should a dashboard have?
There is no official number. Aim for what fits on one screen and answers the dashboard's main question, usually a few cards plus a few charts.
How do I practise building dashboards?
Use a small dataset where you know the right totals, such as the one in the DAX practice exercises, and rebuild the Brightline overview above. Then compare your cards with the numbers in this post.
Practise this on a realistic dataset
Solve it yourself on one company's data, with instant feedback that points at what's wrong without giving away the answer.
- Audit: average order value by segmentAudit
- Completed orders by quarter
- Revenue against target by quarter