BI Developer Quickstart — Your First Report in 30 Minutes¶

Estimated time: 30 minutes | Difficulty: Beginner | What you'll build: A Power BI report connected to the CSA Gold layer through a Direct Lake semantic model, with three interactive visuals — a bar chart, a KPI card, and a detail table — published to the Power BI Service for team consumption.

Prerequisites¶

Before you begin, make sure the following are in place:

• Power BI Desktop installed (latest monthly release recommended)
• Microsoft Fabric capacity (F2 or higher) or an active Fabric trial — Direct Lake requires Fabric
• Gold layer data available — the dbt pipeline must have run at least once (see QUICKSTART.md)
• Workspace member or Contributor role on the target Fabric workspace
• Network access to the Power BI Service and OneLake endpoints

Architecture¶

graph LR
    subgraph "CSA Gold Layer"
        DIM1[dim_customer]
        DIM2[dim_product]
        DIM3[dim_date]
        FACT[fact_sales]
    end

    subgraph "OneLake"
        LH[Lakehouse<br/>Delta Tables]
    end

    subgraph "Fabric Workspace"
        SM[Direct Lake<br/>Semantic Model]
        RPT[Power BI Report]
    end

    subgraph "Power BI Service"
        PUB[Published Report]
        APP[App / Dashboard]
    end

    DIM1 --> LH
    DIM2 --> LH
    DIM3 --> LH
    FACT --> LH

    LH --> SM --> RPT --> PUB --> APP

    style LH fill:#e8f5e9,stroke:#2e7d32
    style SM fill:#e3f2fd,stroke:#1565c0
    style RPT fill:#fff3e0,stroke:#e65100
    style PUB fill:#f3e5f5,stroke:#6a1b9a

Data flows left to right: Gold-layer Delta tables land in a Fabric Lakehouse via OneLake. A Direct Lake semantic model reads columns directly from Parquet files -- no data copy, no scheduled refresh. Reports consume the semantic model and are shared through the Power BI Service.

Step 1 — Access the Fabric Workspace¶

1. Open https://app.fabric.microsoft.com and sign in with your organizational account.
2. In the left navigation, select Workspaces and locate the CSA-in-a-Box workspace (or the workspace your admin has designated for BI development).
3. Verify that the workspace is assigned to a Fabric capacity (F2+) or a trial capacity. The capacity name appears under the workspace settings gear icon.

No Fabric capacity? You can start a 60-day Fabric trial from the Account Manager in the top-right corner of the portal. Direct Lake is not available on shared (Pro-only) capacity.

Step 2 — Create a Lakehouse and Verify Gold Tables¶

If a lakehouse already exists with the Gold tables, skip to Step 3.

1. In the workspace, click + New item and select Lakehouse.
2. Name it csa_gold_lakehouse and click Create.
3. In the lakehouse explorer, click Get data > New shortcut to create shortcuts to the Gold-layer Delta tables in ADLS Gen2.
4. For each table (fact_sales, dim_customer, dim_product, dim_date), create an Azure Data Lake Storage Gen2 shortcut:
   • URL: https://<storage-account>.dfs.core.windows.net
   • Container: gold
   • Sub-path: the folder name matching the table
5. After creating the shortcuts, verify that all four tables appear under the Tables section of the lakehouse.

Tables/
  dim_customer
  dim_date
  dim_product
  fact_sales

Step 3 — Create a Direct Lake Semantic Model¶

1. From the lakehouse view, click New semantic model in the top ribbon.
2. Name the model CSA Sales Model.
3. Select all four tables: fact_sales, dim_customer, dim_product, dim_date.
4. Click Confirm. Fabric creates the semantic model and opens the model view.

Define Relationships¶

In the model diagram, create the following relationships (Fabric may auto-detect some):

Drag from the fact column to the dimension column to create each relationship. Verify cardinality is set to Many-to-One and cross-filter direction is Single.

Step 4 — Define DAX Measures¶

Open the semantic model and select the fact_sales table. Add the following measures using New measure in the ribbon.

Measure 1 — Total Revenue¶

Total Revenue =
SUM( fact_sales[total_amount] )

Format as Currency with 2 decimal places.

Measure 2 — YoY Growth %¶

YoY Growth % =
VAR _CurrentYear = [Total Revenue]
VAR _PriorYear =
    CALCULATE(
        [Total Revenue],
        SAMEPERIODLASTYEAR( dim_date[full_date] )
    )
RETURN
    IF(
        NOT ISBLANK( _PriorYear ),
        DIVIDE( _CurrentYear - _PriorYear, _PriorYear )
    )

Format as Percentage with 1 decimal place.

Measure 3 — Running Total¶

Running Total =
CALCULATE(
    [Total Revenue],
    FILTER(
        ALL( dim_date[full_date] ),
        dim_date[full_date] <= MAX( dim_date[full_date] )
    )
)

Format as Currency with 2 decimal places.

Measure 4 — Average Order Value¶

Avg Order Value =
DIVIDE(
    [Total Revenue],
    DISTINCTCOUNT( fact_sales[sale_id] )
)

Format as Currency with 2 decimal places.

Step 5 — Build the Report in Power BI Desktop¶

Connect to the Semantic Model¶

1. Open Power BI Desktop.
2. Select Home > Get Data > Power BI semantic models (or Dataverse > Power BI semantic models).
3. Choose CSA Sales Model from the workspace and click Connect.
4. Power BI Desktop opens in Live Connection mode — all tables and measures from the semantic model are available in the Fields pane.

Visual 1 — Revenue by Product Category (Bar Chart)¶

1. From the Visualizations pane, select Clustered bar chart.
2. Drag dim_product[category_name] to the Y-axis.
3. Drag Total Revenue to the X-axis.
4. Sort descending by Total Revenue (click the ellipsis on the visual, Sort axis > Total Revenue, then Sort descending).

Visual 2 — KPI Card¶

1. Select Card from the Visualizations pane.
2. Drag Total Revenue into the Fields well.
3. Duplicate the card (Ctrl+D) and replace the field with YoY Growth %.
4. Duplicate again and use Avg Order Value.
5. Arrange the three cards in a row at the top of the report canvas.

Visual 3 — Sales Detail Table¶

1. Select Table from the Visualizations pane.
2. Add the following columns:
   • dim_date[full_date]
   • dim_customer[customer_name]
   • dim_product[product_name]
   • fact_sales[quantity]
   • Total Revenue
3. Set conditional formatting on Total Revenue — select the column, open Format > Cell elements > Background color, and apply a gradient from white to your accent color.

Step 6 — Apply Formatting and CSA Theme¶

Apply a consistent visual style across the report.

1. Go to View > Themes > Browse for themes.
2. If your team has a CSA theme JSON file, import it. Otherwise, create a minimal theme:

{
    "name": "CSA-in-a-Box",
    "dataColors": [
        "#1565C0",
        "#2E7D32",
        "#E65100",
        "#6A1B9A",
        "#C62828",
        "#00838F"
    ],
    "background": "#FFFFFF",
    "foreground": "#212121",
    "tableAccent": "#1565C0",
    "visualStyles": {
        "*": {
            "*": {
                "title": [
                    {
                        "fontFamily": "Segoe UI Semibold",
                        "fontSize": 12
                    }
                ]
            }
        }
    }
}

1. Save this as csa-theme.json and import it via Browse for themes.
2. Adjust the report title: insert a text box at the top with the text CSA-in-a-Box Sales Dashboard using your theme's primary color.

Step 7 — Publish to Power BI Service and Share¶

1. In Power BI Desktop, click Home > Publish.
2. Select the CSA-in-a-Box workspace and click Select.
3. Wait for the publish confirmation dialog, then click Open in Power BI to view the report in the browser.
4. To share with colleagues:
   • Quick share: Click Share on the report toolbar and enter email addresses.
   • App: In the workspace, click Create app to bundle the report into a distributable Power BI App with its own permissions boundary.
   • Teams: Click Chat in Teams or embed the report in a Teams tab for real-time collaboration.

Performance Tips¶

Direct Lake Fallback¶

Direct Lake can fall back to DirectQuery if a query exceeds engine limits. Monitor fallback events in the Capacity Metrics app.

Visual Count and Aggregations¶

Keep visuals per page under 15-20 to avoid slow page loads and capacity throttling. For very large fact tables, build pre-aggregated Gold tables in dbt (for example, fact_sales_monthly_agg) and add them to the semantic model to reduce query volume.

Refresh Behavior¶

Direct Lake models do not need a scheduled refresh. When Delta tables are updated (after a dbt run), the model frames the new data automatically on the next query, or you can trigger framing via the XMLA endpoint:

Invoke-ProcessTable -Server "powerbi://api.powerbi.com/v1.0/myorg/<workspace>" `
    -Database "CSA Sales Model" -Table "fact_sales" -RefreshType "Automatic"

Troubleshooting¶

What's Next¶

• Power BI deep dive -- Connectivity modes, RLS, paginated reports: guides/power-bi.md
• Fabric migration roadmap -- Transitioning to Direct Lake: patterns/power-bi-fabric-roadmap.md
• Fabric end-to-end example -- Lakehouse-to-report walkthrough: examples/fabric-e2e.md
• Data governance -- Sensitivity labels, Purview, RLS: tutorials/02-data-governance