I'll analyze the customer_churn_prediction_yhf data and generate a comprehensive PPT report. Let me start by loading the relevant skills and exploring the data.
I have the schema. This is a customer churn prediction dataset with dimensions like Churn Status, Contract Type, Payment Method, and measures like Churn Rate, Monthly Charges, Support Calls, Tenure, etc. Let me now load the semantic SQL and slide references, then start exploring the data.
Excellent! I have a clear picture of the data and the pipeline. Let me now explore the data thoroughly in Phase 1.
The dataset is customer_churn_prediction_yhf with:
- Dimensions: Churn Status, Contract Type, Customer ID, Monthly Charges, Payment Method, Support Calls, Tenure (Months), Total Usage (GB)
- Measures: Average Monthly Charges, Average Support Calls, Average Tenure, Churn Count, Churn Rate, Unique Customers, Total Customers, Total Monthly Charges, Total Usage (GB)
Let me start the analysis
Excellent data! Now I have a comprehensive picture. The churn rate is very high (86.2%) in this dataset. Let me analyze the key findings and then generate the charts.
Key findings from exploration:
- Overall: 500 customers, 431 churned (86.2% churn rate)
- Contract Type: Monthly contracts have the highest churn rate (90.6%), Two-Year lowest (84.8%)
- Payment Method: Bank Transfer highest churn (89.6%), Credit Card lowest (83.9%)
- Tenure: 7-12 months has highest churn rate (96.2%), 25+ months lowest (76.0%)
- Support Calls: Higher support calls = higher churn - 0 calls = 67.9% churn, 3+ calls = 97%+
- Monthly Charges: Higher charges = higher churn (Very High >$90: 98.2%)
- Usage: Medium usage (101-300GB) has highest churn (87.8%)
- Churned customers: Higher avg charges ($63.5 vs $43.4), shorter tenure (17.7 vs 26.1 months), more support calls (1.63 vs 0.58)
Now let me generate the charts for the PPT
I see the issue — the SQL alias for record_count needs to match the Cube member short name exactly. Let me fix the column aliases.
All 7 charts generated successfully. Now I need to write the summary and outline, then proceed with the PPT pipeline. Let me write the summary and outline files
Now I need to proceed with the presentation pipeline. Let me ask the user for template selection.
The user selected the "clarity" template. Let me proceed with the report pipeline
The guideline mapped all 11 pages with 7 charts successfully. Now let me generate the presentation
The presentation was generated successfully. Let me deliver it with result_reporter and `present_files
Bayes:
✅ Customer Churn Risk Analysis Report — Complete
Your Customer_Churn_Risk_Analysis.pres has been generated and delivered — 11 slides covering the full churn analysis.
Report Structure
| Page | Topic |
|---|---|
| 1 | Cover — Customer Churn Risk Analysis Report |
| 2 | Executive Summary — Key findings at a glance |
| 3 | Overall Churn Profile — Churned vs Retained metrics comparison |
| 4 | Churn by Contract Type — Monthly contracts dominate churn volume |
| 5 | Churn by Payment Method — Bank Transfer highest risk |
| 6 | Support Calls — The strongest churn predictor (2+ calls = 97%+ churn) |
| 7 | Monthly Charges — Strong positive correlation with churn |
| 8 | Tenure & Lifecycle — First 12 months are most critical |
| 9 | Usage Volume — Not a strong differentiator |
| 10 | Key Risk Segments — Highest-risk customer profile |
| 11 | Strategic Recommendations — 5 actionable retention strategies |
Key Findings
- 86.2% churn rate — 431 of 500 customers predicted to churn
- Churned customers: $63.5 avg charges, 17.7 months tenure, 1.63 support calls
- Retained customers: $43.4 avg charges, 26.1 months tenure, 0.58 support calls
- Top risk indicators: 2+ support calls, Monthly contracts, $61+ charges, <12 months tenure
Let me load the dashboard skill to build an interactive BI dashboard.
Now let me load the design template and the reference files
Now let me design the dashboard. I need to:
- Check what explores are available
- Look at existing charts from the analysis that I can reuse
- Create new charts needed for the dashboard
- Build the dashboard JSON
Let me first check the explores and explore the cube
Now I have the explore. I need to create charts for the dashboard. I'll reuse the existing analysis charts where possible, and create new KPI charts. Let me check what chart files already exist on disk
Let me write the dashboard design doc first, then create the charts and dashboard. I'll reuse the existing analysis charts where possible and create new KPI charts via save_embed_chart.
Now let me create the KPI charts using save_embed_chart, then build the dashboard JSON. Let me first check what existing chart files exist.
All KPI charts created. Now let me check what chart JSON files exist to build the dashboard
All charts are in place. Now let me examine the existing chart JSON files to understand their structure for the dashboard JSON
Now I have the complete picture of the chart JSON structure. Let me also check the other charts to understand the full set of field IDs
Now I have a full understanding of the chart JSON structure. Let me also check the remaining charts to ensure completeness
Now I have all chart details. Let me build the dashboard JSON. I need to use the proper field IDs from the explore
Now let me verify the dashboard
Dashboard verification passed! Let me deliver it
Bayes:
✅ Customer Churn Risk Dashboard — Complete
The interactive dashboard customer_churn_dashboard has been built and verified successfully. Here's what's included:
Dashboard Layout (4 sections, 18 tiles)
1. KPI Hero Band — 4 big-number KPIs at the top
| KPI | Value |
|---|---|
| Churn Rate | 86.2% |
| Total Customers | 500 |
| Churned Customers | 431 |
| Average Monthly Charges | $60.70 |
2. Contract & Payment Analysis (2 charts)
- Churn by Contract Type — Monthly contracts highest volume (211 of 233 churned)
- Churn by Payment Method — Bank Transfer highest risk (89.6% churn rate)
3. Service & Charge Analysis (2 charts)
- Churn by Support Calls — 2+ calls = 97%+ churn rate (strongest predictor)
- Churn by Monthly Charges — Very High (>$90) = 98% churn rate
4. Customer Lifecycle & Profile (2 charts)
- Churn by Tenure — First 12 months highest risk, 25+ months most stable
- Average Metrics Comparison — Churned vs retained: charges, tenure, support calls
Interactive Filters
Users can filter the entire dashboard by:
- Churn Status (Churned / Retained)
- Contract Type (Monthly / Annual / Two-Year)
- Payment Method (Credit / Debit / Bank Transfer)
The dashboard is available as dashboards/customer_churn_dashboard.json and ready to embed.
