Deep Data Agent
Tips & Examples

Tips & Examples

Get the most out of Deep Data Agent with these practical tips and real-world examples.

Effective Prompting

Be Specific

Instead of:

“Analyze the data”

Try:

“Calculate the month-over-month growth rate for each product category and highlight any categories with negative growth”

Provide Context

Instead of:

“Show me the top customers”

Try:

“Show me the top 10 customers by total order value in 2024, including their email and number of orders”

Request Specific Formats

Instead of:

“Show sales data”

Try:

“Create a bar chart showing quarterly sales by region, with values in millions”

Example Workflows

Sales Analysis 

1. "Load the sales_2024.csv file and show me a summary"

2. "What's the total revenue by region?"

3. "Show that as a bar chart with the highest region highlighted"

4. "Which products are underperforming compared to last year?"

5. "Create a report with the top insights and export as PDF"

Customer Segmentation 

1. "Connect to the customers database"

2. "How many customers do we have by signup year?"

3. "Segment customers by total purchase value: high (>$1000), medium ($100-1000), low (<$100)"

4. "What's the average order frequency for each segment?"

5. "Create a visualization showing segment distribution"

Data Cleaning Pipeline 

1. "Load raw_data.csv and show me data quality issues"

2. "How many rows have missing values in each column?"

3. "Fill missing numeric values with the column median"

4. "Remove duplicate rows based on customer_id and order_date"

5. "Export the cleaned data as cleaned_data.parquet"

Code Mode Examples

Data Exploration 

Show me the first 10 rows and data types for orders.csv
What's the correlation between price and quantity?
Find outliers in the revenue column using IQR method

Visualization 

Create a scatter plot of orders by date, colored by status
Build a dashboard with:
- Monthly revenue trend
- Top 10 products pie chart
- Orders by region heatmap
Show me a box plot of order values grouped by customer segment

Data Transformation 

Create a pivot table: rows=region, columns=quarter, values=sum of revenue
Add a column for days_since_last_order for each customer
Normalize the price column to a 0-100 scale

SQL Mode Examples

Basic Queries 

How many orders were placed last month?
What's our best-selling product by quantity?
Show customers who haven't ordered in 90 days

Complex Analysis 

Calculate the 7-day moving average of daily revenue
Find customers whose order value increased by more than 50% year-over-year
What's the customer lifetime value distribution by acquisition channel?

Joins and Aggregations 

Show me product performance with category names and supplier info
Rank regions by both revenue and order count
Compare this month's performance to the same month last year

Common Patterns

Iterative Analysis

Start broad, then narrow down:

  1. “Show me an overview of the data”
  2. “Focus on the revenue trends”
  3. “Break it down by region”
  4. “Why is the West region declining?”

Comparative Analysis 

Compare sales performance:
- This year vs last year
- By region
- Show percentage change

Time Series 

Show monthly trends for the past 12 months with:
- Actual values
- 3-month moving average
- Year-over-year comparison

Cohort Analysis 

Group customers by signup month and show:
- Retention rate by month
- Average order value over time
- Cumulative revenue

Pro Tips

1. Use Variables

Reference previous results:

Using that filtered data, calculate the average

2. Chain Operations

Build on previous work:

Now exclude the outliers and recalculate

3. Ask for Explanations

Understand the approach:

Explain your methodology for this analysis

4. Request Alternatives

Get multiple perspectives:

Show me two different ways to visualize this

5. Save Your Work

Export for later:

Save this analysis as a Python script I can reuse

Troubleshooting Examples

When Results Look Wrong 

That doesn't look right. Show me the SQL/code you used
I think you're missing the WHERE clause for active customers only

When You Need More Detail 

Break that down by month instead of quarter
Show me the individual records, not just the aggregate

When Analysis Is Too Slow 

Use a sample of 10,000 rows for this exploration
Can you optimize that query? It's taking too long
RAG Features