Skip to content
Tutorials Center
My learning

Topics

Lesson 2 of 56 ยท Welcome to Power BI

What Is Power BI?

Learn what Power BI is, how Desktop, the online service and mobile apps fit together, and what semantic models, reports and dashboards are.

Power BI is Microsoft's business intelligence tool. It takes data from files, databases and online services, lets you clean and model it, and turns it into interactive reports that people can filter, click through and share. Companies use it to answer questions like "which region is behind plan?", "what did we sell last quarter compared with the quarter before?" or "which customers have stopped ordering?".

The name covers several products that work together, and beginners often mix them up. In this lesson you will learn what each piece does, what the words semantic model, report and dashboard mean, who uses Power BI, and how licences work in broad terms. The whole track builds on this picture, so it is worth getting it right early.

Power BI is a point-and-click application. It cannot run inside this website, so every step you meet in this track is described in words, and every piece of Power Query or DAX code is shown in read-only blocks. To let you experiment anyway, most lessons also contain runnable Python stand-ins (pandas) that reproduce what Power BI computes. Each lesson says which part is the real tool and which part is the stand-in.

The three parts of Power BI

One brand, three places to work

Power BI Desktop A free Windows app where you build.
Power BI service The website where you share and refresh.
Mobile apps View reports on a phone or tablet.

You build in Desktop, publish to the service, and read anywhere.

1 / 2

Power BI Desktop is the authoring tool. It is free to download and runs on Windows only. Here you connect to data, clean it with Power Query, build the data model, write DAX measures and design report pages. You save your work as a .pbix file. A Mac user needs a virtual machine or the browser-based editor in the service, which can do some but not all of what Desktop does.

The Power BI service (also called Power BI online, at app.powerbi.com) is the cloud side. You publish from Desktop into a workspace, and then colleagues open the report in a browser. The service also refreshes data on a schedule, controls who may see what, lets people subscribe to a report by email, and hosts dashboards.

The mobile apps for iOS, Android and Windows show the same reports and dashboards from the service, with touch-friendly navigation. You can also design a separate phone layout for a report.

Semantic model, report and dashboard

Three nouns come up constantly, and they are different things.

Term What it is Where you make it
Semantic model The prepared data: tables, relationships, measures, security rules. Formerly called a dataset. Desktop (Power Query, Model view, DAX)
Report One or more pages of visuals built on one semantic model. Desktop or the service
Dashboard A single canvas of tiles pinned from one or more reports. It exists only in the service. The service

Older articles and menus still say "dataset". Microsoft renamed it to semantic model in late 2023, and you will see both words. They mean the same object.

The key idea is that a report is only a view of a model. The model holds the numbers and the business logic ("revenue means quantity times price, minus returns"). The report decides which parts to show. Several reports can read one model, so a definition such as "revenue" is written once and every report agrees.

flowchart LR
    A["Data sources<br/>files, databases, web"] --> B["Power Query<br/>clean and shape"]
    B --> C["Semantic model<br/>tables, relationships, DAX"]
    C --> D["Report<br/>pages and visuals"]
    D --> E["Dashboard<br/>pinned tiles"]

Data flows left to right; each step builds on the one before it.

From data to insight, with pandas

You cannot run Power BI here, but you can reproduce its flow in pandas. The next example mirrors the stages in the diagram: raw rows (a data source), cleaning (Power Query), a lookup table and a join (the model), a calculation (a DAX measure) and a summary (a visual). Notice how the stages map to Power BI features, not just the numbers.

Python Stand-in: the data-to-insight flow

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

In real Power BI each stage would be a feature: Get Data and the Power Query Editor for steps 1 and 2, Model view for step 3, a measure for step 4 and a bar chart for step 5. The numbers come from pandas; the workflow is the real one.

One model, many reports

Because a model is separate from a report, you can slice the same numbers in different ways without redefining anything. The stand-in below defines one "measure" as a function and uses it for two "reports" that only differ in their filter.

Python Stand-in: one model, two reports

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Change total_sales once and both reports change. That is the same benefit you get from a measure in a shared semantic model.

Who uses Power BI

Power BI serves several roles, and the roles shape which parts of this track matter most to you.

  • Report creators and analysts build reports in Desktop. They spend most of their time in Power Query, DAX and visual design.
  • Data modellers and BI developers design the semantic models that many reports share, and care about performance, security and refresh.
  • Report consumers open published reports in a browser or phone. They filter and explore but do not edit.
  • Administrators manage workspaces, gateways, tenant settings and governance.

You do not need to be a programmer. Power Query is mostly clicks that record steps for you, and DAX is a formula language closer to Excel than to a general-purpose language. Knowing SQL or Excel formulas helps, and if you know pandas you will recognise many of the ideas.

Licences in broad strokes

Licensing changes from time to time, so check Microsoft's current pricing page before you decide. The broad structure is stable:

  • Power BI Desktop is free and has no time limit.
  • A Free licence in the service lets you work in your personal space, but not share your content with others.
  • Pro lets you publish to shared workspaces and share with other people who also have Pro (or who view content on a larger capacity).
  • Premium Per User (PPU) adds larger models, more frequent refresh and advanced features, per person.
  • Capacity-based licensing (Premium capacity, or Fabric capacities) gives your organisation dedicated resources. On a large enough capacity, people with only a Free licence can view shared content.

In practice: learning on your own costs nothing, because Desktop is free and you can sign up for a developer or trial account to try the service. Sharing inside a company usually needs Pro licences or a capacity.

Treat the service as a separate step you learn later. Desktop alone is enough to follow the first five chapters of this track.

Installing Power BI Desktop

Power BI Desktop runs on Windows (64-bit Windows 10 or 11). There is no native Mac version: Mac users work in the browser (the service lets you create and edit reports, with fewer features), or run Windows in a virtual machine or a cloud PC. There are two ways to install it on Windows:

  • Microsoft Store. Search for Power BI Desktop and click Get. The Store version updates itself and does not need administrator rights in most setups. It is the easiest choice for a personal machine.
  • Installer from Microsoft's download page. You download the installer (.exe) and run it. Updates are manual: you download the new installer. Companies often choose this route, because IT can control the version and deploy it centrally.

Microsoft releases a new Desktop version every month, so update regularly: new features, and the screens in this track, are based on a recent version. After installing, sign in with a work or school account (a personal Gmail or Outlook address is not accepted for the service), or skip signing in: Desktop works fully offline for building reports. You can sign up for a Microsoft 365 developer or trial account to try the service without paying.

Choosing a license

Licensing decides who can publish, who can view, and how big your models can be. The main options are:

License Who it is for Can publish and share? Typical limits
Free Learning, personal use Not to shared workspaces Work in "My workspace" only
Pro Authors and viewers in a team Yes, and viewers also need Pro Models up to about 1 GB; up to 8 scheduled refreshes a day
Premium Per User (PPU) Teams that need advanced features without a whole capacity Yes, with other PPU users Models of up to about 100 GB; up to 48 refreshes a day
Fabric / Premium capacity Organizations with many viewers or large models Authors may need Pro; viewers can have a Free license on a large enough capacity Large models, more refreshes, advanced features

The numbers are approximations from Microsoft's documentation at the time of writing and they change, so check the current Power BI pricing and licensing pages before you decide.

The consequences matter more than the details:

  • Everyone who opens shared content needs a suitable license: with Pro it is every viewer, with a large capacity it is only the authors. This is the main cost driver for large audiences.
  • If you start on Free, you can build and keep everything in Desktop, and only the sharing step requires an upgrade.
  • If viewers are external (customers or partners), sharing options and licensing are different again: look at Embedded or guest access with your administrator.
  • Features such as paginated reports, deployment pipelines and larger models depend on the license or capacity.
Python Stand-in: a license chooser

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The function is only a mental model: real decisions also depend on price lists, existing Microsoft 365 agreements and which features you need.

Power BI compared with Tableau

Many people ask which tool to learn. Both are mature products and both are widely used, so the honest answer is "it depends on where you work". Here is a fair summary:

Topic Power BI Tableau
Starting cost Desktop is free; sharing needs Pro or a capacity Authoring and viewing need paid licenses
Microsoft and Excel fit Very strong (Excel, Teams, SharePoint, Azure, Fabric) Works with Microsoft data but is not tied to it
Modelling and calculations Full data model with relationships, DAX measures and Power Query Strong calculated fields and level-of-detail expressions; the model is lighter
Visual freedom Good, with custom visuals Very flexible, famous for visual exploration
Platform Desktop on Windows only Desktop on Windows and Mac
Community and jobs Very large in Microsoft-based organizations Large, especially in design-oriented analytics teams

The skills transfer: data modelling, picking the right chart and thinking about filters apply to both. Learn the one your target employer or current company uses, and the second will be easy to pick up.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Where do you normally build a Power BI report, including the data model and DAX measures?

    Power BI Desktop is the authoring tool for Power Query, the model, DAX and report pages. The service is mainly for sharing, refresh and security, and the mobile apps are for viewing.
  2. 2. What is a semantic model?

    A semantic model (formerly called a dataset) holds tables, relationships, measures and rules. Reports are views on top of it.
  3. 3. Which statement about dashboards is true?

    Dashboards live in the service and collect tiles pinned from one or more reports. Reports can have several pages and are built on a single model.
  4. 4. Why is it useful that several reports can share one semantic model?

    A shared model keeps business logic in one place. Change the measure once and every report that uses it shows the new result.

Summary

  • Power BI has three main parts: Desktop (free, Windows, for authoring), the service (cloud, for sharing and refresh) and mobile apps (for viewing).
  • A semantic model (formerly dataset) holds the data, relationships, measures and security; a report is a set of visuals built on one model; a dashboard is a canvas of pinned tiles in the service.
  • The flow is data sources, Power Query, model, report, dashboard.
  • Desktop is free; sharing in an organisation normally needs Pro, Premium Per User or a capacity licence.
  • Python stand-ins in this track reproduce the numbers; the real tool is point-and-click.

Next, take a guided look around the application in A Tour of Power BI Desktop.