Publishing & Power BI Service

Definition

The Power BI Service (app.powerbi.com) is the cloud platform where reports built in Desktop are published, shared, refreshed on a schedule, and consumed by other users - organized into workspaces, with content distributed to end users via Apps.


Publishing a Report

Power BI Desktop: Home > Publish > choose a destination workspace

Publishing uploads the report AND creates/updates its underlying dataset

A .pbix file published to the Service becomes both a Report (the visual pages) and a Dataset (the data model + DAX) - these can later be managed somewhat independently (e.g. multiple reports can be built against one shared published dataset).


Workspaces

graph TD
    A[Workspace] --> B[Datasets]
    A --> C[Reports]
    A --> D[Dashboards]
    A --> E[Dataflows]
    A --> F[Members & Access]
Power BI Service > Workspaces > + New workspace
Workspace roleCan do
Adminfull control - manage members, delete workspace, all content permissions
Memberedit/publish content, manage some settings
Contributoredit/publish content, cannot manage workspace settings/members
Viewerview content only, no editing

Workspaces are the unit of collaboration

Everyone with at least Contributor access to a workspace can see and edit its content - separate workspaces per team/department (rather than one giant shared workspace) keeps permissions manageable and reduces accidental overwrites.


Apps (Distributing to Broad Audiences)

Workspace > Create app > select which reports/dashboards to include > Publish app

Apps vs sharing a workspace directly

An App is a curated, read-only, polished bundle of specific reports/dashboards from a workspace, distributed to end users (who never see the underlying workspace, its editing tools, or in-progress content) - the standard way to deliver finished reports to a broad, non-technical audience, as opposed to giving them direct workspace access.


Scheduled Refresh

Workspace > dataset > "..." > Settings > Scheduled refresh
Refresh frequency: Daily / Weekly
Time(s) of day: e.g. 6:00 AM, 6:00 PM
Time zone
Failure notification email

Scheduled refresh requires stored credentials for every data source

Set these under Dataset Settings > Data source credentials - a refresh silently fails (or emails a failure notice) if credentials expire (e.g. a password change) or were never configured for a newly added source.

TierMax scheduled refreshes/day
Free / Pro8 per day
Premium / Fabric capacity48 per day

Gateways (On-Premises Data Sources)

graph LR
    A["On-premises database
(e.g. internal SQL Server)"] --> B["On-premises Data Gateway
(installed locally)"]
    B -->|encrypted connection| C["Power BI Service
(cloud)"]

Why a gateway is needed

The Power BI Service runs in Microsoft’s cloud and has no direct network access to a database sitting inside a company’s private network - the On-premises Data Gateway (a small installed application) acts as a secure bridge, so scheduled refresh can reach internal data sources without exposing them directly to the internet.

Download: Power BI > Download > On-premises data gateway
Register the gateway with your organization's Power BI tenant
Workspace > Dataset Settings > Gateway connection > select the registered gateway

Personal Gateway vs Standard/Enterprise Gateway

A Personal Gateway is simple but only works for that one user’s own scheduled refreshes and can’t be shared. A Standard/Enterprise Gateway (installed on a dedicated always-on machine/server) supports multiple data sources and multiple users’ datasets - the correct choice for any real team/organizational use.


Dataflows (Reusable, Centralized Power Query)

Workspace > + New > Dataflow > define entities using Power Query Online

Dataflows centralize ETL logic across multiple reports

Instead of repeating the same Power Query transformations inside every individual .pbix file, a Dataflow defines the transformation ONCE in the Service, and multiple datasets/reports can then connect to that shared, pre-cleaned output - reduces duplicated logic and inconsistency across a report portfolio.


Dashboards vs Reports

ReportDashboard
Sourceone dataset, multiple pagescan pin tiles from MULTIPLE reports/datasets onto one canvas
Interactivityfull (filters, slicers, drillthrough)limited - mostly click-through to the source report
Best forin-depth explorationat-a-glance summary combining several sources
Open a visual in a report > pin icon (thumbtack) > Pin to a dashboard

Sharing Options

Share button on a report:
  Share (direct link, specific people/groups)
  Publish to web (PUBLIC, embeds anywhere - use with extreme caution for sensitive data)
  Embed in SharePoint / Teams / a custom app (via Power BI Embedded)

"Publish to web" makes a report PUBLICLY accessible to anyone with the link

No login or organizational access control applies - never use this option for reports containing sensitive, internal, or personally identifiable data. Reserve it strictly for genuinely public data intended for open sharing.


Version History & .pbix Management

Workspace > report > "..." > Version History      - restore a previous published version if needed

Keep source .pbix files in a proper version control location too

Power BI Service version history is convenient but limited - many teams also store the actual .pbix files in SharePoint/OneDrive/git-based storage (via tools like pbi-tools for git-friendly diffing) for more robust backup and collaboration history.