Power BI is three separate pieces that work together: Desktop, a free Windows app for building a data model and a report; the Service, the cloud site at app.powerbi.com where a published report is stored, viewed and shared; and Mobile, a phone and tablet app for checking an already-published report on the go. The normal workflow runs in that order: build in Desktop, publish to a workspace in the Service, then view or share from the Service or Mobile.
Below is what each piece actually does, how a dataset differs from a report and a dashboard, how Import mode differs from DirectQuery, and the mistakes that trip up a beginner, or an interview candidate, most often.
In this guide
- The three pieces: Desktop, Service, Mobile
- The basic workflow, in order
- Inside Desktop: Report, Data and Model view
- Dataset versus report versus dashboard
- Import mode, DirectQuery and Live Connection
- Publishing, workspaces, refresh and gateways
- Common mistakes and misconceptions
- Cheat sheet
- Try it yourself
- FAQ
The three pieces: Desktop, Service, Mobile
Power BI is not one single application. It is a free authoring tool, a cloud site, and a lightweight viewer, and each one has a different job.
| Piece | What it is | What you do there |
|---|---|---|
| Power BI Desktop | A free Windows application | Connect to data, shape it in Power Query, build the data model and relationships, write DAX measures, and design every report page and visual |
| Power BI Service | The cloud site, app.powerbi.com | View, share and organise a published report into workspaces, schedule a data refresh, pin tiles onto a dashboard, and set up row-level security |
| Power BI Mobile | A phone and tablet app | View an already-published report or dashboard, drill into a visual, set a data alert, or leave an annotation, while away from a computer |
A simple way to remember it: Desktop is where things get built, the Service is where things get published and viewed, and Mobile is where things get checked on the go. Building a brand new report is not something either the Service or Mobile is for; both are built around consuming a report that Desktop already produced.
The basic workflow, in order
Short version
Build the model and report in Desktop → publish it to a workspace in the Service → view, share or pin it from the Service, or check it from Mobile. Authoring always starts on the desktop application, never in the cloud.
Every Power BI project, however small, follows the same four steps:
- Connect and shape data in Desktop. Get Data pulls in a source, Power Query’s Transform Data cleans and reshapes it, and the result loads into the model.
- Build the model and the report, still in Desktop. Relationships are set up in Model view, DAX measures are written, and the report’s pages and visuals are built in Report view.
- Publish to a workspace in the Service. One click sends the dataset and the report from the local .pbix file up into a workspace the author picks.
- View, share or pin, from the Service or Mobile. Colleagues open the report in a browser, a scheduled refresh keeps an Import-mode dataset current, and individual visuals can be pinned as tiles onto a dashboard for an at-a-glance summary.
Publishing is not a one-way door. The author keeps editing the local .pbix file in Desktop as usual, and clicking Publish again simply overwrites the earlier published version in that workspace with the newer one. Editing the report directly inside the Service, on the other hand, does not update the local file back in Desktop: once published, the Service copy and the local file are two separate things, each edited independently.
Inside Desktop: Report, Data and Model view
Desktop has three main views, switched with the icons down the left edge, and each shows a completely different side of the same file.
| View | What it shows | Typical use |
|---|---|---|
| Report view | The actual report pages and visuals, with the Fields pane and the Visualizations pane | Dragging fields onto a page to build and arrange charts, tables and slicers |
| Data view | The loaded rows of a selected table, much like a plain spreadsheet grid | Spot-checking values, or adding a calculated column |
| Model view | A diagram of every table in the model and the relationships connecting them | Creating, inspecting or adjusting a relationship, and setting its cardinality and filter direction |
The Fields pane, visible in Report view, lists the tables, columns and measures of the loaded model, which is exactly what gets dragged onto a visual to build it. It is the model’s own table of contents, not a list of report pages or colours.
Dataset versus report versus dashboard
These three words get used loosely in conversation, but inside Power BI they are three distinct objects, and an interview question built around the difference comes up often.
| Object | Built where | What it actually is |
|---|---|---|
| Dataset | Desktop (the model behind a .pbix) | The data, the model of tables and relationships, and the measures, kept as its own object separate from any report |
| Report | Desktop, or the Service | Multiple interactive pages built against one dataset, with full cross-filtering between visuals |
| Dashboard | The Service only | A single page of tiles pinned from one or more reports, mostly static, and not something Desktop builds directly |
Two details worth remembering: a single dataset can sit underneath more than one report, so several teams can each build their own report against the one shared dataset without duplicating data; and a single dashboard can combine pinned tiles from several different reports and datasets onto one page, which is exactly what makes a dashboard useful as a cross-cutting summary. The key practical difference between a report and a dashboard is that a report keeps its full page structure and interactivity, while a dashboard flattens selected tiles onto a single, simpler summary page.
A .pbix file, the file Desktop saves, is a self-contained bundle of all of it: the compressed data, the Power Query steps, every DAX measure, and the report pages with their visuals all travel together inside that one file.
Import mode, DirectQuery and Live Connection
How a model connects to its source changes how fresh the numbers are, and how fast the report feels, and it is one of the most common things an interviewer probes.
| Mode | What happens | Trade-off |
|---|---|---|
| Import | A compressed copy of the data is loaded into the file itself | Fast, since every visual is answered locally, but it is a snapshot that only updates at the next refresh |
| DirectQuery | No data is copied in; every visual sends its own live query back to the source | Always current, but speed depends entirely on how fast the source itself responds |
| Live Connection | The report connects directly to an existing external model, most often another published dataset, without importing or building a new one | Reuses a model that already exists elsewhere, rather than owning its own copy of the data |
Import feels more responsive because the data already sits inside Power BI’s own fast, compressed engine, so a filter click is answered locally instead of waiting on a round trip to an external source. DirectQuery trades that speed for currency: since nothing is stored locally, each click that changes a filter triggers a fresh query against the real source, so a slow source produces a visibly slow report no matter how well the visuals are designed.
A model does not have to pick only one mode for every table. A composite model mixes Import and DirectQuery within the same file, for example importing a small, slow-changing lookup table for speed while leaving a huge, constantly growing fact table on DirectQuery rather than copying all of it in.
Refresh only matters for Import mode. Refreshing an Import-mode dataset re-runs the source queries and reloads a fresh copy of the data, replacing the previous snapshot, which is why a report viewed right after its source changes, but before a refresh runs, still shows the older snapshot. A DirectQuery report has no local snapshot to go stale in the first place, so it needs no equivalent scheduled refresh.
Publishing, workspaces, refresh and gateways
Once a report is ready in Desktop, four Service-side concepts take over.
- Publish sends the dataset and report from the local .pbix file up into a workspace in the Service, which is what makes the content available for others to view.
- A workspace is a container in the Service that holds a related set of datasets, reports and dashboards together, typically everything belonging to one team or project.
- Scheduled refresh re-runs an Import-mode dataset’s source queries on a schedule, so viewers see data that is current as of the last scheduled run rather than the moment the file was first published.
- A gateway lets the cloud-based Service securely reach a data source that sits on a local network or an on-premises machine, which a scheduled refresh needs whenever the source itself is not already reachable from the public internet.
Power BI also has a natural-language Q&A box: typing a plain question, such as total sales by region last year, automatically builds a matching visual from the model without manually dragging fields.
Common mistakes and misconceptions
These are the traps that show up most often, in real use and in interview questions alike.
- Assuming Import mode is always live. It is not. Import is a snapshot; a change at the source is invisible until a refresh actually runs.
- Assuming a published report stays linked to the local file. It does not. After publishing, the Service copy and the local .pbix are independent, so an edit in one never automatically updates the other.
- Thinking DirectQuery needs the same scheduled refresh as Import. It does not, since there is no local snapshot sitting around to go stale.
- Confusing a dashboard with a report. A dashboard is a single page of pinned tiles built in the Service; a report is a multi-page, fully interactive object built in Desktop.
- Treating Mobile as an authoring tool. It is a viewing and light-interaction surface, not a place new reports get built.
Cheat sheet
| Question | Answer |
|---|---|
| Where is a report built? | Power BI Desktop |
| Where is a report viewed and shared? | Power BI Service (app.powerbi.com) |
| Where is a report checked on the go? | Power BI Mobile |
| What is a dataset? | The data, model and measures, separate from any report |
| What is a dashboard built in? | The Service, by pinning tiles |
| Which mode needs a scheduled refresh? | Import mode only |
| Which mode is always current? | DirectQuery |
| What mixes Import and DirectQuery in one model? | A composite model |
| What lets the Service reach an on-premises source? | A gateway |
Try it yourself
Quick check
A colleague’s report is viewed the moment after its source data changes, with no refresh run yet. If the dataset is Import mode, does the report show the new values or the old ones? What if it were DirectQuery instead?
Show answer
Import mode shows the older snapshot from the last refresh, since nothing updates a loaded dataset until a new refresh actually runs. DirectQuery would show the new values immediately, since every visual queries the live source at the moment it is viewed.
Frequently asked questions
What is the difference between Power BI Desktop and Power BI Service?
Desktop is the free Windows application used to build a data model and a report. The Service is the cloud site, app.powerbi.com, where a published report is stored, viewed, shared and scheduled to refresh.
Is Power BI Desktop free?
Yes. Desktop itself is free to download and use for building a model and a report; a licence generally comes into play for publishing to the Service and sharing with others, depending on the organisation’s setup.
What is the difference between a Power BI report and a dashboard?
A report has multiple interactive pages with full cross-filtering and is built in Desktop or the Service. A dashboard is a single page of tiles pinned from one or more reports, built only in the Service, and is mostly static.
What is the difference between Import mode and DirectQuery in Power BI?
Import mode copies a compressed snapshot of the data into the file, which is fast but only updates at the next refresh. DirectQuery copies nothing in and sends a live query to the source every time a visual is viewed, so it is always current but depends on the source’s own speed.
Can Power BI Mobile be used to build a new report?
No. Mobile is for viewing an already-published report or dashboard, drilling into a visual, or setting a data alert, not for authoring new report pages.
How often does a Power BI Desktop file update?
Desktop has long followed a roughly monthly update cadence, regularly prompting the user to install a newer build with new features and fixes.
Test yourself
Timed questions on Power BI Basics and Workflow, with an explanation for every answer.