ProPrompt for Analysts

Target Audience: Business Analysts, Data Analysts, BI Specialists, Controllers, and anyone working with data, reports, and insights.

Getting Started – Your First Analysis Prompt
Summarizing & Preparing Data
Generating Reports & Dashboards
Advanced – Complex Analyses
Agent: Automated Analysis Pipelines
Cheat Sheet for Analysts

1 Getting Started – Your First Analysis Prompt

Difficulty: ⭐ Easy

First things first: A good analysis prompt follows the RICE framework (→ see Fundamentals).

Example – Simple Data Summary

You are an experienced Business Analyst.

Summarize the following sales data:
- Period: Q1 2026
- Products: Licenses, Support Contracts, Training
- Regions: DACH, Nordics, UK

Output the summary as a Markdown table with:
| Region | Product | Revenue | Change vs. Previous Quarter |

Why does this work? Role, context, task, and format are clearly defined.

Tips for Beginners

Tip	Description
🎯 Be specific	“Analyze Q1 revenue by region” instead of “Look at the data”
📊 Define format	Table, bullet points, or JSON – always specify
🔢 Name your KPIs	Which metrics? Revenue, margin, growth, churn?
📅 Define the period	Always state the analysis time frame

2 Summarizing & Preparing Data

Difficulty: ⭐⭐ Medium

Example – Preparing Excel Data for AI

You are a data analyst. I have an Excel file with customer data.

The columns are:
| Column | Type | Description |
|--------|------|-------------|
| customer_id | INT | Unique customer ID |
| revenue_2025 | FLOAT | Annual revenue 2025 |
| segment | STRING | Enterprise / SMB / Startup |
| churn_risk | FLOAT | Churn probability (0-1) |

Task:
1. Create a segment analysis with average revenue per segment
2. Identify the top 10 customers by revenue with high churn risk (> 0.7)
3. Output everything as Markdown tables

Example – SQL Query from Natural Language

You are a BI analyst with SQL expertise.

Database: PostgreSQL
Tables:
- orders (id, customer_id, amount, order_date, status)
- customers (id, name, segment, region)
- products (id, name, category, price)

Create a SQL query that shows:
- Monthly revenue per region for 2025
- Only completed orders (status = 'completed')
- Sorted by region and month
- With month-over-month comparison (% change)

Output the query with comments.

Extracting Data from Office Files

For converting Excel, Word, or PowerPoint into AI-friendly formats → see Preparing Office Files.

3 Generating Reports & Dashboards

Difficulty: ⭐⭐ Medium

Example – Creating a Management Report

You are a Senior Business Analyst creating a management report.

## Context
- Company: SaaS platform with 2,000 customers
- Period: Q1 2026
- Audience: C-Level and Board

## Data
- MRR: €850,000 (previous quarter: €790,000)
- Churn Rate: 3.2% (previous quarter: 4.1%)
- NPS: 72 (previous quarter: 68)
- New Customers: 145 (previous quarter: 120)

## Task
Create an Executive Summary Report with:
1. **Headline KPIs** as a table with trend arrows (↑/↓/→)
2. **Key Insights** – 3-5 bullet points
3. **Risks & Opportunities** – 2-3 each
4. **Recommended Actions** – prioritized

## Format
- Maximum 1 page
- Professional tone
- English

Example – Visualization Template (Mermaid Diagram)

Create a Mermaid diagram showing monthly revenue trends.

Data:
- Jan: 280k, Feb: 285k, Mar: 285k
- Apr: 290k, May: 295k, Jun: 310k

Use an xychart-beta bar chart with labeled axes.

Result:

xychart-beta
    title "Monthly Revenue 2026 (in kEUR)"
    x-axis [Jan, Feb, Mar, Apr, May, Jun]
    y-axis "Revenue (kEUR)" 250 --> 320
    bar [280, 285, 285, 290, 295, 310]

Example – Process Flow Visualization

flowchart LR
    A[📥 Raw Data] --> B[🔄 Cleansing]
    B --> C[📊 Analysis]
    C --> D{Result OK?}
    D -- Yes --> E[📋 Report]
    D -- No --> B
    E --> F[📧 Send to Stakeholders]

    style A fill:#e1f5fe
    style E fill:#c8e6c9
    style F fill:#fff9c4

4 Advanced – Complex Analyses

Difficulty: ⭐⭐⭐ Hard

Example – Cohort Analysis with Python

You are a Senior Data Analyst with Python/Pandas expertise.

## Goal
Create a cohort analysis for customer retention.

## Data
CSV file with transactions:
- customer_id, order_date, amount

## Requirements
1. Group customers by signup month (cohort)
2. Calculate retention rate for 12 months
3. Create a heatmap with Seaborn
4. Export results as a Markdown table

## Constraints
- Python 3.11, Pandas 2.x, Seaborn
- Only customers with at least 1 order
- Cohort format: YYYY-MM

Example – Describing a Forecasting Model

You are a Data Scientist.

Walk me through building a simple revenue forecasting model step by step:

1. Data preparation (which features?)
2. Model selection (why which model?)
3. Training & validation
4. Interpreting results

Context:
- Monthly revenue data for the last 3 years
- Seasonal fluctuations present
- Python + scikit-learn

Output the code with detailed comments.

5 Agent: Automated Analysis Pipelines

Difficulty: ⭐⭐⭐ Hard

What Is an Analysis Agent?

An agent can autonomously perform multi-step analyses:

Read and clean data
Run calculations
Create visualizations
Generate reports

Example – Reporting Agent (Copilot Studio)

# Role
You are ReportBot, the automated reporting assistant for the Controlling team.

# Capabilities
- You analyze SharePoint lists with financial data
- You create monthly KPI reports
- You compare actuals vs. plan values
- You identify deviations > 10%

# Behavior
- Respond in English
- Use tables and KPI cards
- Round numbers commercially to 2 decimal places
- Use €-format for currencies

# Data Sources
- SharePoint List: "Finance_KPIs_2026"
- SharePoint List: "Budget_Plan_2026"

# Workflow
1. User requests a report (e.g., "Show me the March monthly report")
2. Load data from both lists
3. Calculate: Actual vs. Plan, Deviation %, Trend
4. Create formatted report
5. Highlight critical deviations (> 10%) in red

# Output Format
## Monthly Report [Month] [Year]

| KPI | Plan | Actual | Deviation | Trend |
|-----|------|--------|-----------|-------|
| Revenue | €X | €Y | Z% | ↑/↓ |

### Critical Deviations
- [KPI]: [Details]

### Recommendations
- [Action 1]
- [Action 2]

Agent Toolchain: Automated Analysis Workflow

flowchart TB
    A[👤 Analyst asks:<br/>'Create Q1 Report'] --> B[🤖 Orchestrator Agent]
    B --> C[📥 Data Agent]
    B --> D[📊 Analysis Agent]
    B --> E[📝 Report Agent]

    C --> |Loads raw data| C1[SharePoint / Excel / DB]
    C --> |Cleans and formats| D
    D --> |Calculates KPIs| D1[Deviations<br/>Trends<br/>Forecasts]
    D --> |Results| E
    E --> |Generates| E1[Executive Summary<br/> Detail Report<br/> Email Draft]

    style A fill:#e3f2fd
    style B fill:#fff3e0
    style C fill:#e8f5e9
    style D fill:#fce4ec
    style E fill:#f3e5f5

Agent Prompt for VS Code (Agent Mode)

## Goal
Create a Python script that generates an automated monthly report.

## Context
- Data source: CSV files in /data/monthly/
- Output: Markdown report in /reports/
- Existing structure: /src/analytics/

## Steps
1. Read all CSV files in /data/monthly/
2. Calculate KPIs: Revenue, Costs, Margin, Customer Count
3. Compare with previous month and same month last year
4. Create Mermaid diagrams for trends
5. Generate a Markdown report with Executive Summary
6. Save to /reports/YYYY-MM-report.md

## Requirements
- Python 3.11, Pandas, no other external dependencies
- Error handling for missing files
- Logging with the logging module
- Type hints for all functions

6 Cheat Sheet for Analysts

Quick Prompt Templates

Task	Prompt Starter
Summarize data	`"Summarize the data in #file as a table with [KPIs]."`
Write SQL	`"Write a SQL query (PostgreSQL) that shows [requirement]."`
Excel formula	`"Create an Excel formula that calculates [calculation]."`
Pivot table	`"Explain how to create a pivot table for [analysis]."`
Spot trends	`"Analyze the trend in the following data: [data]"`
Write report	`"Create an Executive Summary report for [audience]."`
Visualization	`"Create a Mermaid diagram showing [data]."`
Find anomalies	`"Identify outliers in the following data: [data]"`

Context Checklist for Analysis Prompts

Data source specified? (CSV, Excel, DB, API)
Columns/fields described?
Time period defined?
KPIs named?
Target audience for output clear? (Management, Team, Stakeholders)
Format specified? (Table, Chart, Report)

Back to overview: 🏠 Home · Fundamentals (DE) · Fundamentals (EN)

Created by Justin Szczepaniak · GitHub Project · LinkedIn

ProPrompt for Analysts

Table of Contents

1 Getting Started – Your First Analysis Prompt

Difficulty: ⭐ Easy

Example – Simple Data Summary

Tips for Beginners

2 Summarizing & Preparing Data

Difficulty: ⭐⭐ Medium

Example – Preparing Excel Data for AI

Example – SQL Query from Natural Language

Extracting Data from Office Files

3 Generating Reports & Dashboards

Difficulty: ⭐⭐ Medium

Example – Creating a Management Report

Example – Visualization Template (Mermaid Diagram)

Example – Process Flow Visualization

4 Advanced – Complex Analyses

Difficulty: ⭐⭐⭐ Hard

Example – Cohort Analysis with Python

Example – Describing a Forecasting Model

5 Agent: Automated Analysis Pipelines

Difficulty: ⭐⭐⭐ Hard

What Is an Analysis Agent?

Example – Reporting Agent (Copilot Studio)

Agent Toolchain: Automated Analysis Workflow

Agent Prompt for VS Code (Agent Mode)

6 Cheat Sheet for Analysts

Quick Prompt Templates

Context Checklist for Analysis Prompts