How to build a Power BI dashboard: a step-by-step guide for beginners

Microsoft describes a Power BI dashboard as a single page, often called a canvas, that tells a story through visuals. Each visual on it is a tile you pin from a report, and selecting a tile takes you to the report behind it.

For a beginner the practical upshot is simple. The real work happens in a report. The dashboard is the front page you assemble afterwards, often from more than one report, for people who want the headline numbers at a glance.

Reports and dashboards look alike, but they behave differently:

Power BI Desktop is a free download from Microsoft, either from the Microsoft Store or as an installer. Microsoft lists Windows 10 or later as the minimum, and there's no Mac version, so on a Mac you'd need Windows running in a virtual machine.

For data, start with something you understand, such as a year of sales with a date, product, region and amount on every row. An export from your accounts software, CRM or online shop is ideal, as long as it has one row per transaction and a heading on every column. If it's in Excel, format it as a table first (Ctrl+T), because Power BI picks up tables cleanly.

Then, before you open anything, answer three questions on paper. Who will read this? What decision will it help them make? And which three to five questions must it answer at a glance? For a sales dashboard, they might be:

In Power BI Desktop, select Get data on the Home ribbon and choose Excel workbook or Text/CSV. Pick your file, tick the table you want in the Navigator, and select Transform data rather than Load. That opens Power Query, where the cleaning happens.

Power Query records every change you make as a step, listed under Applied steps. When next month's file arrives, a refresh replays the same steps on the new data, so you clean it once rather than every month. That alone is worth the effort of learning it.

For a first dashboard, these fixes cover most messy exports:

If all your data sits in one table, you can skip most of the modelling for now. A single, well-cleaned table makes a perfectly good first dashboard.

Once you add a second table, such as a product list with categories, set things up as a star schema. It's a plain idea with a grand name: one central table of transactions (the fact table) surrounded by tables that describe them (dimension tables), such as products, customers, regions and dates. In Model view you connect them with relationships, usually one product to many sales. A relationship does the job a VLOOKUP or XLOOKUP would do in Excel, without copying a single column across.

Then write a measure. Measures are calculations written in DAX, Power BI's formula language, and they recalculate for whatever the reader has selected. Select New measure on the Home ribbon and type:

Total Sales = SUM(Sales[Amount])

Two more give you the core of almost any sales dashboard. Orders = COUNTROWS(Sales) counts the rows, and Average Order Value = DIVIDE([Total Sales], [Orders]) divides one by the other. DIVIDE is better than a slash because it returns a blank rather than an error when there's nothing to divide by.

You could drag the Amount column straight onto a chart and let Power BI add it up for you. Writing measures instead means every number is defined once, named clearly and reused everywhere, so your cards and charts can't quietly disagree.

Go back to the questions you wrote down and give each one a visual. If a chart doesn't answer one of them, it doesn't go on the page.

Then lay the page out the way people read: headline cards across the top, the main trend underneath, comparisons beside it and detail at the bottom. Keep the slicers together in one place. Give every visual a title that says what it shows, such as "Sales by month, last 12 months", rather than the default field names.

Go easy on colour. One main colour for normal and one accent for what needs attention makes the important thing stand out, while a different colour for every bar makes nothing stand out. Skip pie charts with more than a few slices, because slices are hard to compare by eye. Our rule of thumb: if the page needs scrolling, it's two pages.

Each chart type has a job:

Save your file, select Publish on the Home ribbon and choose a workspace. Signing in to the Power BI service needs a work or school account, because Microsoft doesn't accept personal addresses such as Gmail or Hotmail. With a free licence you can publish to My workspace and build dashboards there for your own use.

Open the report in the service, hover over a visual and select the pin icon. Choose New dashboard, give it a name, and that visual becomes your first tile. Repeat for each visual you want on the front page. If you'd rather keep the page interactive, so selecting one chart filters the others, pin the entire report page instead (from the More options menu) and it comes across as one live tile.

Sharing is where licences come in. Microsoft's guide to Power BI licence types sets it out: a free licence covers content you create for yourself, sharing needs Power BI Pro or Premium Per User, and in most organisations the people viewing it need a paid licence too. The exception is content saved in a workspace on Premium or Fabric F64 (or larger) capacity, which colleagues on free licences can view. Check what your organisation has before you promise your manager a link.

Last, sort out refresh, so the numbers update without you republishing the file every week.

Almost every first dashboard goes wrong in the same few places. Knowing them in advance saves an afternoon of confusion:

Excel is better when the data is small, lives in one workbook, and the dashboard is for you or a few colleagues who all have Excel. Pivot tables, pivot charts and slicers make a perfectly good one-page dashboard, and nobody needs a new tool or a new licence.

Power BI is better when the data is large or comes from several sources, when it refreshes every week or month, and when lots of people need to read it in a browser or on their phone. An Excel worksheet stops at 1,048,576 rows, and a workbook passed around by email has a habit of becoming five slightly different workbooks by Friday. A published Power BI report is one version that everyone opens.

The good news is that the two are closer than they look. Excel has Power Query built in, the same cleaning tool Power BI uses, and Excel's Power Pivot uses the same DAX formula language. Learn those in Excel and Power BI will feel familiar from day one, which is why our Certificate in Excel for Data Analysis is a sensible first step if Power BI feels like a leap.

Build one, then build another with different data. The steps above stay the same every time, and the second dashboard always comes together faster than the first.

Here's the honest part: you may not need a course at all. Microsoft's own Power BI training on Microsoft Learn is free, self-paced and genuinely good, and it's the route towards Microsoft's own certification, exam PL-300. If you're disciplined and only need the tool, start there.

A paid course earns its keep when you want more than the tool: a structured project that takes you from a raw file to a published report, help when you're stuck, and something to show for it. Our Certificate in Power BI is built that way. You finish with a portfolio dashboard built from a real dataset, the OBA Tutor is on hand when you get stuck, and you earn a certificate employers can verify online. It's an OBA certificate of completion rather than a Microsoft certification, and the dashboard you build is the stronger proof anyway.

If dashboards are part of a bigger plan, read our guide on how to become a data analyst and look at the Data Analyst Bundle, which pairs Power BI with Excel for Data Analysis and SQL. If your workplace runs Tableau instead, the same planning and design rules apply, and the Certificate in Tableau teaches that tool. Not sure yet? The free info pack sets out what each course covers before you spend a penny.