BD-Sales Pairing Optimization Analysis

A comprehensive data analytics project analyzing Business Development and Sales representative pairing performance to optimize lead routing and maximize conversion rates and revenue.

🌐 Live Interactive Dashboards

👉 View Live Dashboards

Explore the complete analysis through three interactive dashboards:

• Dashboard 1: Exploratory Data Analysis & Data Validation
• Dashboard 2: Performance Methodology & Analysis Results
• Dashboard 3: Business Recommendations & $600K+ ARR Impact

📊 Project Overview

This project analyzes 2,200+ opportunities across 18 BD reps and 23 Sales reps (413 unique pairings) to identify optimal BD-Sales team pairings and provide data-driven routing recommendations.

Business Problem

Not all BD-Sales pairings perform equally. Random lead routing leads to:

• Suboptimal win rates
• Longer sales cycles
• Higher early deal death rates
• Inconsistent pipeline health

Solution

Analyze historical pairing performance to:

1. Identify high-performing BD-Sales combinations
2. Flag at-risk pairings to avoid
3. Provide actionable routing recommendations
4. Quantify expected business impact in ARR

🎯 Key Findings

Expected ARR Improvement: $600K+ (annual recurring revenue lift from optimized routing)

• 18 BDs analyzed with varying pairing performance ranges
• Top 3 BDs show $100K+ ARR improvement potential each
• 124-point performance swing between best and worst routing decisions for some BDs
• $0 implementation cost - pure routing logic optimization

📁 Project Structure

sales-handoff-quality-analysis/ │ ├── data/ │ └── opportunities.csv # Simulated opportunity data │ ├── analysis/ │ ├── pair_metrics.csv # BD-Sales pair metrics │ ├── performance_scores.csv # Performance scoring results │ ├── bd_pairing_recommendations.csv # Top/bottom 5 recommendations │ └── routing_impact_analysis.csv # ARR impact calculations │ ├── visualizations/ │ ├── output/ # Generated PNG visualizations │ └── *.py # Visualization scripts │ ├── iterations/ # Project evolution │ ├── v1_initial_approach.py │ ├── v2_added_confidence.py │ ├── v3_baseline_comparison.py │ └── notes.txt │ ├── exploratory_data_analysis.py # EDA and data validation ├── data_generation.py # Simulated data generation ├── metric_calculation.py # Calculate pairing metrics ├── performance_scoring.py # Score and classify pairings │ ├── dashboard_index.html # Interactive dashboard landing page ├── dashboard_1_eda.html # EDA dashboard ├── dashboard_2_methodology.html # Methodology & analysis dashboard ├── dashboard_3_recommendations.html # Business recommendations dashboard │ ├── METHODOLOGY.md # Detailed methodology doc ├── README.md # Project documentation ├── requirements.txt # Python dependencies └── .gitignore # Git ignore file

Technologies Used

• Python 3.8+: Core analysis language
• pandas: Data manipulation and analysis
• numpy: Numerical computing
• matplotlib: Data visualization
• seaborn: Statistical visualizations
• HTML/CSS: Interactive dashboards

Getting Started

1. Clone the repository

git clone https://github.com/yourusername/sales-handoff-quality-analysis.git cd sales-handoff-quality-analysis

2. Install dependencies pip install -r requirements.txt

3. Run the analysis pipeline

Step 1: Generate simulated data

python data_generation.py

Step 2: Exploratory data analysis

python exploratory_data_analysis.py

Step 3: Calculate metrics

python metric_calculation.py

Step 4: Score performance

python performance_scoring.py

Step 5: Generate visualizations

python visualizations/run_all_visualizations.py

4. View the dashboard

📈 Analysis Methodology

1. Data Collection

• Simulated 2,200 opportunities spanning 2 years
• 18 BD reps × 23 Sales reps = 413 possible pairings
• Tracked: Win rate, early death rate, stale pipeline rate, deal size

2. Metric Calculation

Four key metrics calculated for each pairing:

• Win Rate: % of decided opportunities that closed won
• Early Death Rate: % of lost deals that died within 14 days
• Stale Pipeline Rate: % of open deals inactive for 90+ days
• Average Deal Size: Mean deal value

3. Performance Scoring

• Baseline Comparison: Each pair compared to that BD's average (controls for lead quality)
• Percentage Deviation: Relative performance vs baseline
• Weighted Scoring: Equal 25% weight per metric
• Confidence Multiplier: Based on opportunity count (threshold: 7)

4. Classification

Percentile-based classification:

• High Performer: Top 25% (75th percentile+)
• Above Average: 50th-75th percentile
• Average: 25th-50th percentile
• Below Average: 10th-25th percentile
• At-Risk: Bottom 10%

📊 Key Visualizations

Dashboard 1: Exploratory Data Analysis

1. Opportunity Distribution: Shows confidence levels across pairings
2. Data Quality Validation: Confirms dataset is ready for analysis

Dashboard 2: Methodology & Performance Analysis

3. Performance Score Distribution: Percentile thresholds and classification zones
4. Top/Bottom Performers: Best and worst BD-Sales combinations
5. Performance Heatmap: Complete pairing matrix
6. Metric Contributions: What drives high vs low performance

Dashboard 3: Business Recommendations

7. BD Pairing Recommendations: Performance range for each BD
8. Routing Decision Matrix: Green (route) vs red (avoid) lookup table
9. ARR Impact Analysis: Quantified revenue improvement from optimization

---

💼 Business Impact

Current State (Random Routing)

• Total ARR: ~$1.2M
• No systematic pairing optimization
• Performance left on the table

Optimized State (Route to Top 5)

• Total ARR: ~$1.8M
• +$600K ARR improvement
• 50% ARR increase

🎓 Skills Demonstrated

Technical Skills

• Data analysis and manipulation (pandas, numpy)
• Statistical analysis and percentile-based classification
• Data visualization (matplotlib, seaborn)
• Python scripting and automation
• HTML/CSS dashboard creation

Business Skills

• Problem identification and scoping
• Metric definition and KPI selection
• Data-driven recommendation development
• Stakeholder communication and storytelling
• ROI analysis and business case building

Analytical Skills

• Baseline methodology selection
• Confidence interval application
• Performance classification systems
• Business impact quantification

🚧 Challenges Encountered

1. Baseline Selection Challenge

Problem: How do we fairly compare pairings when different BDs pass different quality leads?

Initial Approach: Compare all pairings to company-wide average

• Issue: Didn't account for BD lead quality differences
• Enterprise BDs naturally have lower win rates than SB BDs

Solution: Compare each pairing to that BD's average (BD baseline)

• Controls for inherent lead quality from each BD
• Makes comparisons fair and actionable for routing decisions

Impact: This was the most critical methodological decision - completely changed the analysis validity

2. Circular Logic in BD Summary

Problem: When averaging all BD pairs' performance scores, results were close to zero

Initial Reaction: Thought there was a bug in the calculation

Realization: This is mathematically correct by design!

• We compare each pair to that BD's average
• Deviations naturally balance out (some above, some below)
• This is a feature, not a bug

Solution:

• Documented this clearly in methodology
• Added explanation box to visualization
• Changed interpretation: "This measures pairing consistency, not absolute BD quality"

Lesson Learned: What looks like an error might be correct behavior - understand your methodology deeply

3. Confidence Threshold Selection

Problem: How many opportunities needed for reliable pairing analysis?

Trial & Error:

• Tried threshold of 3 opps → Too noisy, unreliable results
• Tried threshold of 10 opps → Lost too many pairings, insufficient data
• Tried threshold of 5 opps → Still some instability

Solution: Two-tier approach

• Minimum threshold: 3 opportunities (43% confidence)
• Full confidence: 7 opportunities (75th percentile in dataset)
• Apply confidence multiplier that scales linearly

Impact: Balances statistical reliability with data availability

4. Visualization Complexity

Problem: Initial visualizations tried to show everything at once - too overwhelming

Examples of cuts made:

• Sales Rep frequency analysis → Too complex for main dashboard
• Confidence vs Performance scatter → Interesting but not essential
• Multiple classification views → Chose most intuitive representation

Solution:

• Focus on storytelling: Each dashboard has a clear narrative
• Progressive disclosure: Start simple (EDA), build to complex (recommendations)
• Keep "cut" visualizations in analysis folder for reference

Lesson Learned: Less is more - every chart must earn its place in the story

5. ARR Calculation Complexity

Problem: How do we estimate revenue impact when not all opportunities are closed?

Initial thought: Only count won deals

• Issue: Ignores the opportunity cost of poor routing

Solution: Calculate potential ARR based on win rate × deal value × volume

• Current: Actual win rate × avg deal × opportunities
• Optimized: Best pairing win rate × avg deal × opportunities
• Difference = ARR improvement potential

Assumption: Average deal size remains constant across pairings (reasonable for this analysis)

Impact: Provides compelling business case in dollar terms

📝 Future Enhancements

• Real-time dashboard with Plotly/Dash
• Machine learning model for pairing prediction
• Time-series analysis of pairing performance trends
• A/B test framework for routing validation
• Integration with CRM systems (Salesforce, HubSpot)
• Automated monitoring and alerting
• Sales rep specialization analysis (which BDs work best with each rep)

---

👤 Author

Ain

Data Analyst | Marketing Analyst | Revenue Operations

• Portfolio: []
• LinkedIn: [(https://www.linkedin.com/in/qurat-ul-ain-236b60133/)]
• GitHub: @Ain-spec
• Email: quratulain_izhar@hotmail.com

Additional Documentation

• METHODOLOGY.md: Detailed technical methodology

Contributing

This is a portfolio project, but feedback and suggestions are welcome! Feel free to:

1. Open an issue for discussion
2. Fork the repo and experiment
3. Share how you've adapted this methodology for your use case

Contact

Have questions about the methodology or want to discuss similar projects?

Reach out via LinkedIn or email

Last Updated: January 2026