This chapter covers the three core components of Microsoft Power BI: Reports, Dashboards, and Datasets. Understanding the differences and relationships between these components is critical for the DP-900 exam, particularly in the Analytics domain (objective 3.4). Approximately 10-15% of exam questions touch on Power BI concepts, and many of these test your ability to distinguish reports from dashboards and understand the role of datasets. We will define each component, explain its purpose and limitations, and show how they interact in a typical Power BI workflow.
Jump to a section
Imagine a large corporation. The CEO needs to see the big picture of company performance. The Dashboard is the CEO's executive dashboard — a single screen with key performance indicators (KPIs) like revenue, profit margin, and employee count. It's designed for at-a-glance monitoring and is updated periodically (e.g., daily). Below the CEO, department heads need deeper analysis. They use Reports — detailed documents with multiple pages, filters, and drill-down capabilities. For example, the sales director has a report showing sales by region, product, and salesperson, with charts and tables. Reports are interactive and can be explored. Underlying everything are the Datasets — the raw data sources. These are like the company's data warehouse: tables of transactions, customer records, and product lists. Datasets are prepared, cleaned, and modeled by data analysts. The CEO's dashboard pulls summarized metrics from a dataset, while the sales director's report queries the same dataset with different filters. The key mechanic: dashboards are limited to one page and are built from visualizations pinned from reports or directly from datasets. Reports can have multiple pages and are built on a single dataset. Datasets are the foundation — they contain the data and relationships. Just as the CEO can't drill into individual sales transactions from the dashboard (they'd need to open the report), a dashboard in Power BI is a lightweight view that redirects to underlying reports for details. The dataset ensures consistency: everyone sees the same numbers, just at different levels of detail.
Power BI is a business analytics service by Microsoft that provides interactive visualizations and business intelligence capabilities. The platform is built around three fundamental objects: Reports, Dashboards, and Datasets. Each serves a distinct purpose and has specific constraints that are frequently tested on the DP-900 exam.
Reports are multi-page, interactive documents that present data from a single dataset. They allow users to explore data through visualizations like charts, tables, maps, and slicers. Reports are created in Power BI Desktop or the Power BI service and can be shared with others. Key characteristics include:
Multiple pages (tabs) with different visualizations on each.
Interactive features: cross-filtering, drill-down, drill-through, and tooltips.
Built on a single dataset (though a dataset can be used by multiple reports).
Can be exported to PDF, PowerPoint, or Excel.
Dashboards are single-page canvases that display a collection of visualizations (called tiles) that can come from multiple reports and datasets. They are designed for at-a-glance monitoring and are limited to one page. Key characteristics include:
Tiles can be pinned from a report, a Q&A question, or an Excel workbook.
Dashboards can contain tiles from different datasets, but each tile is based on a single visualization.
They support natural language queries (Q&A) and can have alerts set on tile data.
Dashboards are read-only; you cannot edit the underlying visualizations directly.
Datasets are the containers for data that is imported or connected to Power BI. They include tables, relationships, measures, and calculated columns. Datasets are the foundation for reports and dashboards. Key characteristics include:
Can be sourced from many data sources: Azure SQL Database, Excel, CSV, SharePoint, etc.
Support import mode (data copied into Power BI) or DirectQuery (live connection to source).
Can be scheduled to refresh at intervals (e.g., daily).
A single dataset can be used by multiple reports and dashboards.
How They Interact
The typical workflow: A data analyst connects to data sources and builds a dataset in Power BI Desktop (or in the service). They then create a report with multiple pages of visualizations based on that dataset. The report is published to the Power BI service. From the report, a user can pin a visualization to a dashboard. The dashboard now shows a tile from that report. The dashboard can also include tiles from other reports or from Q&A. When a user clicks a tile on the dashboard, they are taken to the underlying report for deeper analysis.
Key Differences Tested on DP-900
The exam frequently asks questions that require you to identify whether a given scenario describes a report or a dashboard. Here are the definitive differences:
Number of pages: Reports can have multiple pages; dashboards have exactly one page.
Data source: A report is based on a single dataset; a dashboard can combine tiles from multiple datasets.
Editing: Reports can be edited (add/remove visualizations) in the service (if you have permissions); dashboards cannot be edited directly — you pin tiles from other objects.
Interactivity: Reports support cross-filtering, drill-down, and drill-through; dashboards support limited interactivity (clicking a tile navigates to the source report, and tiles can be filtered by clicking on them if they are from the same dataset).
Alerts: Dashboards can have alerts set on tiles (e.g., email when a metric exceeds a threshold); reports do not support alerts.
Natural language Q&A: Dashboards have a Q&A box at the top; reports do not have this feature.
Creating and Managing Reports
Reports are created in Power BI Desktop or in the Power BI service using the web-based editor. In Desktop, you connect to data, build a data model (dataset), and then design visualizations on report pages. Once published, the report is available in the service. You can also create reports directly in the service by using an existing dataset.
Key actions in the service: - Edit report: Opens the report in editing mode where you can add, remove, or modify visualizations. - Pin to dashboard: Select a visualization and pin it to a dashboard (existing or new). - Export: Export report data to CSV or export the whole report to PDF. - Subscribe: Subscribe to receive email notifications of the report on a schedule.
Creating and Managing Dashboards
Dashboards are created only in the Power BI service. You cannot create a dashboard in Power BI Desktop. To create a dashboard: 1. Click the "+ New dashboard" button in the service. 2. Give it a name. 3. Pin tiles from reports, Q&A, or other sources.
Key actions in the service: - Add tile: You can add tiles from reports, Q&A, or custom web content. - Edit tile: Change the title, subtitle, or hyperlink of a tile. - Set alert: On a tile, you can set an alert rule (e.g., notify when value exceeds a threshold). - Share: Share the dashboard with other users or groups. - Feature as featured: Set a dashboard as the featured dashboard for a user.
Datasets in Detail
Datasets are the data models that Power BI uses. They contain: - Tables: Imported or connected data tables. - Relationships: Connections between tables (e.g., foreign key relationships). - Measures: DAX formulas for aggregations (e.g., SUM, AVERAGE). - Calculated columns: Columns defined by DAX formulas. - Hierarchies: User-defined hierarchies for drill-down (e.g., Year > Quarter > Month).
Data connectivity modes: - Import: Data is copied into Power BI's in-memory cache. Supports large datasets (up to 1 GB per dataset in shared capacity, 10 GB in Premium). Refreshes can be scheduled. - DirectQuery: No data is imported; queries are sent directly to the source. Suitable for real-time data but limited performance. - Composite models: Combine Import and DirectQuery.
Dataset refresh: - Scheduled refresh: Set up to 8 times per day in shared capacity, 48 in Premium. - Refresh interval: Minimum 30 minutes. - Gateway required for on-premises data sources.
Common Exam Scenarios
Scenario 1: A manager wants a single-page view of KPIs from multiple reports. Solution: Create a dashboard and pin the relevant visualizations from each report.
Scenario 2: An analyst needs to explore data interactively with drill-down and cross-filtering. Solution: Create a report with multiple pages.
Scenario 3: A team needs to share a view that updates automatically and sends alerts when a metric changes. Solution: Create a dashboard with alerts on tiles.
Limitations and Constraints
Dashboard tile count: Maximum 100 tiles per dashboard (50 in some older versions).
Report page count: No hard limit, but performance degrades with too many pages.
Dataset size: In shared capacity, 1 GB limit per dataset; Premium up to 10 GB (or larger with large dataset storage format).
Refresh frequency: Shared capacity 8 times per day; Premium up to 48.
Dashboard Q&A: Only available on dashboards, not reports.
How They Interact with Other Power BI Features
Power BI Apps: Packages of dashboards and reports shared with users. An app can contain multiple dashboards and reports.
Power BI Embedded: Embed reports and dashboards into custom applications.
Power BI Mobile: View dashboards and reports on mobile devices. Dashboards are optimized for mobile viewing.
Power BI Service vs Desktop: Desktop is for authoring reports and datasets; Service is for sharing and consuming (and limited report editing).
Exam Tips
Remember: Dashboards are always one page; reports can have many.
A dashboard tile can come from a report or Q&A; it cannot be created from scratch in the dashboard (except for text boxes, images, and web content).
When you see "single page view" and "from multiple reports/datasets", think dashboard.
When you see "interactive exploration" and "multiple pages", think report.
Datasets are the underlying data model; they are not directly viewed but are the source for reports and dashboards.
The Q&A feature is only on dashboards.
Alerts are only on dashboard tiles.
To share a report, you share the report itself or include it in an app. Dashboards can be shared directly.
Summary of Relationships
A report is based on exactly one dataset.
A dashboard can contain tiles from multiple reports and datasets.
A dataset can be used by multiple reports and dashboards.
A report can be published to a workspace, and from there its visualizations can be pinned to dashboards.
A dashboard is always created in the Power BI service, never in Desktop.
Now you have a thorough understanding of the differences and relationships between Power BI reports, dashboards, and datasets. This knowledge will help you answer exam questions accurately and apply Power BI effectively in real-world scenarios.
Connect to Data Source
Start Power BI Desktop and select 'Get Data'. Choose a data source such as Azure SQL Database, Excel, or CSV. Provide connection details (server, database, credentials). Power BI will either import the data (Import mode) or create a live connection (DirectQuery). For import, a copy of the data is stored in Power BI's internal engine (VertiPaq). For DirectQuery, queries are sent to the source each time a visualization is rendered. The choice affects performance and freshness. The DP-900 exam expects you to know the difference between Import and DirectQuery modes.
Build the Dataset
After connecting, you may transform data using Power Query Editor (e.g., remove columns, change data types, merge tables). Then define relationships between tables (e.g., Sales table linked to Product table via ProductID). Create measures using DAX (e.g., Total Sales = SUM(Sales[Amount])). This dataset becomes the foundation for reports. In the service, datasets can be scheduled to refresh (default: daily, but configurable). The dataset size limit in shared capacity is 1 GB; Premium allows up to 10 GB. The exam may test dataset refresh limits.
Create a Report
In Power BI Desktop, add visualizations to a report page by dragging fields from the dataset onto the canvas. Common visualizations include bar charts, line charts, maps, and tables. Add slicers for filtering. Add additional pages (tabs) for different views. Reports support drill-down (hierarchies), drill-through (navigate to detail page), and cross-filtering (clicking a bar filters other visuals). Reports are interactive and can be exported. Reports are published to the Power BI service.
Publish to Power BI Service
Click 'Publish' in Power BI Desktop. Select a workspace (e.g., 'My Workspace' or a shared workspace). The report and its dataset are uploaded to the service. The dataset becomes a separate entity that can be used by other reports. The report appears in the workspace list. From here, you can share the report, create a dashboard, or set up scheduled refresh. Publishing is a one-way process; to update, you republish from Desktop. The exam may ask about publishing workflow.
Create a Dashboard from the Report
In the Power BI service, open the published report. Hover over a visualization and click the pin icon. Select an existing dashboard or create a new one. The visualization becomes a tile on the dashboard. Repeat for other visualizations from the same or different reports. Dashboards can also include tiles from Q&A (natural language queries) or from Excel workbooks. Dashboards are limited to one page. Clicking a tile navigates to the underlying report. The exam tests that dashboards are created by pinning, not by direct editing.
Configure Dashboard Features
Once the dashboard is created, you can set alerts on tiles (e.g., email when a KPI exceeds a threshold). Use the Q&A box at the top to ask natural language questions (e.g., 'total sales by product'). The dashboard can be shared with other users or groups. You can set a dashboard as featured (appears first when users sign in). Dashboards are read-only; you cannot edit visualizations directly. If you need to change a visualization, you must edit the source report and repin. The exam emphasizes that dashboards are for monitoring, not authoring.
Enterprise Scenario 1: Sales Performance Monitoring
A global retail company uses Power BI to monitor daily sales performance. The sales team has a report with multiple pages: one for sales by region, one for product category performance, and one for individual salesperson rankings. The report uses a single dataset that combines data from their ERP system (Azure SQL Database) and CRM (Salesforce). The dataset is scheduled to refresh every 30 minutes (the maximum allowed in shared capacity is 8 times per day, but they use Premium capacity for 48 refreshes). The regional managers have a dashboard that shows key metrics: total revenue, top-selling products, and sales target achievement. These tiles are pinned from the report. The dashboard also includes a tile from Q&A that answers 'What is the current month's revenue?' The CEO has a separate dashboard that combines tiles from multiple reports (sales, inventory, and finance). In production, the dataset size is around 500 MB, which is within the 1 GB shared limit. Common issues: when the dataset refresh fails due to gateway connectivity, the dashboard shows stale data. The solution is to set up alerts on the dataset refresh status.
Enterprise Scenario 2: Executive Dashboard for KPIs
A financial services firm needs a single view of key performance indicators (KPIs) from different departments. They create a dashboard with tiles from three different reports: one from finance (profit margin), one from operations (customer satisfaction score), and one from HR (employee turnover). Each report is based on its own dataset. The dashboard is shared with the executive team. The dashboard uses the Q&A feature to allow executives to ask ad-hoc questions like 'What was the revenue last quarter?' The dataset for the finance report is in DirectQuery mode to ensure real-time data from their Azure Analysis Services model. The dashboard has alerts on the profit margin tile: if it drops below 20%, an email is sent to the CFO. In production, the dashboard has 30 tiles (well below the 100-tile limit). The main challenge is that the Q&A feature sometimes misinterprets queries, so they train users on specific phrasing. Misconfiguration example: if the dashboard tiles are from different datasets, clicking a tile does not cross-filter other tiles because cross-filtering only works within the same dataset. Users often expect cross-filtering across tiles, but that is not supported.
Enterprise Scenario 3: Operational Report for Data Analysts
A healthcare organization uses Power BI for operational reporting. Data analysts create reports with multiple pages that allow drill-down from hospital-level to department-level to individual patient metrics. The dataset is imported from an on-premises SQL Server database via a gateway. The report uses row-level security (RLS) to restrict data based on user roles (e.g., a hospital manager can only see their hospital's data). The report is published to a workspace and shared with analysts. They use the report's cross-filtering and drill-through features to investigate anomalies. The dataset is refreshed nightly. In production, the dataset size is 800 MB (close to the 1 GB limit). They plan to move to Premium to accommodate growth. Common mistake: analysts try to create a dashboard from the report but expect it to have the same interactivity as the report. They need to understand that dashboards are for monitoring, not deep analysis. The exam tests that RLS applies to datasets, so both reports and dashboards respect the security settings.
DP-900 Exam Focus on Power BI Components (Objective 3.4)
The DP-900 exam tests your ability to differentiate between Power BI reports, dashboards, and datasets. You will not need to create them, but you must understand their characteristics, relationships, and use cases. Specific objectives under 'Analytics' include:
Describe the capabilities of Power BI reports.
Describe the capabilities of Power BI dashboards.
Describe the relationship between datasets, reports, and dashboards.
Identify when to use a report vs. a dashboard.
Common Wrong Answers and Why Candidates Choose Them
'Dashboards can have multiple pages' — This is the most common trap. Candidates confuse dashboards with reports. Remember: dashboards are always single-page. Reports have multiple pages.
'Reports can combine data from multiple datasets' — A report is based on a single dataset. Dashboards can combine tiles from multiple datasets. Candidates often mix this up.
'You can edit visualizations directly on a dashboard' — Dashboards are read-only. To change a visualization, you must edit the source report and repin. Candidates think they can edit tiles like in a report.
'Alerts can be set on reports' — Alerts are only available on dashboard tiles. Reports do not have alerting capability.
'Q&A is available in reports' — Q&A (natural language query) is only available on dashboards, not reports.
Specific Numbers and Terms That Appear on the Exam
Maximum tiles per dashboard: 100 (sometimes 50 in older versions). The exam may ask about this limit.
Dataset refresh frequency: Up to 8 times per day in shared capacity; 48 in Premium. Minimum interval 30 minutes.
Dataset size limit: 1 GB per dataset in shared capacity; 10 GB in Premium (or larger with large dataset storage format).
Pin to dashboard: The action of adding a visualization from a report to a dashboard.
Workspace: A container for reports, dashboards, and datasets.
Power BI Desktop: Tool for creating reports and datasets.
Power BI service: Online platform for sharing and consuming reports and dashboards.
Edge Cases and Exceptions
Dashboards can contain tiles from Excel workbooks uploaded to Power BI, not just from reports.
Dashboards can have tiles from custom streaming datasets (real-time data).
A dashboard can be created from a report by pinning an entire report page (pins all visuals at once).
Reports can be created directly in the service using an existing dataset, without using Desktop.
Datasets can be shared across workspaces (if the user has permission).
How to Eliminate Wrong Answers Using the Underlying Mechanism
When you see a question that asks 'Which Power BI component should you use to...?', follow this logic:
If the scenario mentions multiple pages, it's a report.
If it mentions single page and combining visuals from different sources, it's a dashboard.
If it mentions creating visualizations or interactive exploration, it's a report.
If it mentions monitoring or alerts, it's a dashboard.
If it mentions underlying data model or refresh schedule, it's a dataset.
For example: 'You need to provide a single-page view of KPIs from sales, marketing, and finance reports. Which should you use?' The answer is a dashboard because it combines tiles from multiple reports on one page. If the question said 'You need to allow users to drill down into sales data by region and product,' that's a report because of interactivity and drill-down.
By understanding these distinctions, you can confidently answer any DP-900 question on this topic.
Reports are multi-page, interactive, and based on a single dataset.
Dashboards are single-page, read-only, and can combine tiles from multiple datasets.
Datasets are the underlying data models that power reports and dashboards.
Alerts and Q&A are features exclusive to dashboards.
Dashboards are created by pinning visualizations from reports or other sources.
A report can be edited in the Power BI service; a dashboard cannot.
Dataset refresh is limited to 8 times per day in shared capacity, 48 in Premium.
These come up on the exam all the time. Here's how to tell them apart.
Report
Multiple pages possible
Based on a single dataset
Supports editing (add/remove visuals)
Supports cross-filtering and drill-down
No alerts or Q&A
Dashboard
Single page only
Can combine tiles from multiple datasets
Read-only; visuals pinned from reports
Limited interactivity (click tile to open report)
Supports alerts and Q&A
Mistake
Dashboards can have multiple pages.
Correct
Dashboards are always single-page. Reports can have multiple pages. This is the most common confusion.
Mistake
A report can use data from multiple datasets.
Correct
A report is based on a single dataset. However, a dashboard can combine tiles from multiple datasets.
Mistake
You can edit visualizations directly on a dashboard.
Correct
Dashboards are read-only. To change a visualization, you must edit the source report and repin the tile.
Mistake
Alerts can be set on reports.
Correct
Alerts are only available on dashboard tiles. Reports do not support alerts.
Mistake
Q&A (natural language query) is available in reports.
Correct
Q&A is only available on dashboards. Reports do not have a Q&A box.
Reveal each answer, then mark whether you got it right. Score 60%+ to unlock the next chapter.
A report is a multi-page, interactive document based on a single dataset that allows editing, cross-filtering, and drill-down. A dashboard is a single-page, read-only collection of tiles that can come from multiple datasets. Dashboards support alerts and Q&A, while reports do not. Reports are created in Power BI Desktop or the service; dashboards are created only in the service by pinning visuals from reports.
No, a dashboard always has exactly one page. If you need multiple pages, you should use a report. This is a key distinction tested on the DP-900 exam.
Dashboards are created in the Power BI service by clicking 'New Dashboard' and then pinning tiles from reports, Q&A, or other sources. You cannot create a dashboard in Power BI Desktop. Tiles are added by pinning visualizations from reports or by using the Q&A box to create a tile.
No, a report is based on a single dataset. However, a dashboard can combine tiles from multiple datasets. This is a common exam trap.
The maximum is 100 tiles per dashboard. In some older versions, it was 50. This limit appears on the exam.
No, alerts are only available on dashboard tiles. To set an alert, you must pin the relevant visualization to a dashboard and then configure the alert on that tile.
Import mode copies data into Power BI's in-memory cache, allowing fast performance but requiring scheduled refreshes. DirectQuery mode sends queries directly to the source, providing real-time data but with performance limitations. Import mode has a 1 GB dataset limit in shared capacity; DirectQuery has no size limit but depends on source performance.
You've just covered Power BI: Reports vs Dashboards vs Datasets — now see how well it sticks with free DP-900 practice questions. Full explanations included, no account needed.
Done with this chapter?