4 The Journey Begins
5 Your First Connection
In this chapter, we’ll connect Claude to the SalesVP model — a real Power BI semantic model containing wine and spirits sales data. You’ll learn how to explore data models, write DAX queries, and get business insights through natural conversation.
- How to discover model structure
- How to ask business questions
- How to write and execute DAX queries
- How to interpret and explain results
6 Step 1: Say Hello to Your Data
Let’s start with a simple connection test. This confirms Claude can reach the Power BI model.
6.0.0.1 Try It: Test the Connection
Copy this prompt into Claude Desktop:
“Can you connect to the SalesVP model and tell me what tables are available?”
Claude will:
- Request permission to use the Power BI MCP Server
- Connect to the SalesVP model
- List all tables in the semantic model
You should see tables like:
- Sales — The main fact table with transactions
- Product — Product dimension with names, categories
- Date — Date dimension for time intelligence
- GrossistMap — Wholesaler/distributor mapping
- And more…
7 Step 2: Explore the Model Structure
Now let’s understand what data we’re working with.
7.0.0.1 Try It: Discover Measures
“What measures are defined in the SalesVP model? Show me the most important ones.”
This reveals the pre-built calculations in the model, such as:
| Measure | Description |
|---|---|
Liter |
Total volume sold in liters |
Flasker |
Total units/bottles sold |
Liters ifjor |
Same period last year volume |
Endring Liter % |
Year-over-year change percentage |
YTD |
Year-to-date volume |
Mkt Share |
Market share calculation |
Understanding existing measures helps you leverage the model’s built-in intelligence rather than recreating calculations from scratch.
8 Step 3: Ask Business Questions
Here’s where the magic happens. Instead of writing complex DAX, just ask in plain English.
8.0.0.1 Try It: Natural Language Query
“What are the top 10 best-selling products by volume (liters)?”
Claude will:
- Understand you want a ranking
- Write the appropriate DAX query
- Execute it against the model
- Present the results in a clear format
8.1 More Example Questions
Try these prompts to explore different aspects of the data:
“How has total liter sales changed compared to last year?”
“Which wholesalers (grossister) have the highest sales volume?”
“Show me the monthly sales trend for 2024”
“What is the market share of 3-liter bag-in-box wines?”
9 Step 4: Understanding DAX Queries
As you work with Claude, you’ll see DAX queries being generated. Let’s understand the basics.
9.1 Anatomy of a DAX Query
Example DAX Query
EVALUATE
TOPN(
10,
ADDCOLUMNS(
VALUES(Product[Product Name]),
"Total Liters", CALCULATE(SUM(Sales[Volum]))
),
[Total Liters], 0
)| Component | Purpose |
|---|---|
EVALUATE |
Required keyword to return a table |
TOPN(10, ...) |
Return only the top 10 rows |
ADDCOLUMNS(...) |
Add calculated columns to a table |
VALUES(...) |
Get unique values from a column |
CALCULATE(...) |
Evaluate an expression with context |
Ask Claude to explain the DAX query it generates. This is a great way to learn DAX!
“Can you explain the DAX query you just wrote? What does each part do?”
10 Step 5: Iterative Analysis
Real analysis is iterative. Start broad, then drill down based on what you find.
10.1 Example Workflow
11 Step 6: Advanced Queries
Once comfortable, try more sophisticated analysis:
11.1 Time Intelligence
“Calculate the rolling 12-month average sales and compare it to the previous 12-month period”
11.2 Filtering & Segmentation
“Show me sales by price segment, but only for products over 1 liter”
11.3 Comparative Analysis
“Compare the top 5 products this year vs last year. Which products are growing fastest?”
11.4 Custom Measures
“Create a DAX measure that calculates the average price per liter”
12 Common Patterns
12.1 Pattern 1: Rankings
"Top/Bottom N by [metric]"“What are the bottom 5 performing products by year-over-year growth?”
12.2 Pattern 2: Comparisons
"Compare [A] to [B] by [metric]"“Compare bag-in-box wines to glass bottles by total volume and growth rate”
12.3 Pattern 3: Trends
"Show the trend of [metric] over [time period]"“Show the weekly sales trend for the past 3 months”
12.4 Pattern 4: Anomalies
"Find unusual patterns in [data]"“Are there any products with unusually high or low sales compared to their historical average?”
13 Best Practices
- Start with overview questions before drilling into details
- Ask Claude to explain queries and results
- Verify unexpected results by asking for the underlying data
- Save useful queries for future reference
- Requesting all raw data — ask for aggregations instead
- Vague questions — be specific about what you want
- Ignoring context — Claude remembers your conversation
- Not validating — spot-check results against known values
14 Hands-On Exercise
Now it’s your turn! Complete these tasks using the SalesVP model:
- List all tables in the model
- Find the total number of products (SKUs)
- Identify the date range of the sales data
- Find the top 10 products by sales volume
- Calculate the year-over-year change for those products
- Identify which product has the highest growth rate
Ask Claude to summarize the key findings in a format suitable for a management presentation.
15 Summary
- Connect Claude to Power BI semantic models
- Explore model structure (tables, measures, relationships)
- Query data using natural language
- Understand generated DAX queries
- Iterate from overview to detailed analysis
16 Next Steps
In Chapter 4: Making Confident Decisions, we’ll learn how to validate your findings with statistical analysis — ensuring you can make data-backed decisions with confidence.