📊 A/B Testing Conversion Analysis | Python | Data Analytics Project

🚀 Executive Summary

This end-to-end analytics project evaluates conversion performance of an old vs. new landing page using Python, pandas, and matplotlib.

The objective is to simulate how a data analyst supports product and marketing teams by analyzing user behavior, calculating conversion rates, performing statistical significance tests (Z-test), and providing actionable recommendations for website optimization.

Workflow:
Raw Data → Data Cleaning → Conversion Rate Analysis → Statistical Testing → Visualization → Actionable Insights

---

🧠 Business Questions Answered

• What is the conversion rate for the old landing page vs. the new landing page?
• Is the difference in conversion statistically significant?
• Should the new landing page be launched or kept in testing?
• How do conversion rates differ visually between the two pages?

---

🛠 Tech Stack

• Python 3.x
• pandas (Data manipulation)
• numpy (Statistical calculations)
• matplotlib (Visualization)
• CSV dataset handling
• Statistical Analysis (Z-test)

---

📂 Repository Structure

ab-testing-conversion-analysis/
│
├── data/
│   └── ab_data.csv
│
├── images/
│   └── conversion_rate_chart.png
│
├── ab_test.py
└── README.md

---

🔍 Analytical Framework

1️⃣ Data Cleaning

• Remove rows where group and landing_page do not match
• Remove duplicate users to prevent double-counting

2️⃣ Conversion Rate Calculation

• Calculate mean conversion rates for control (old page) and treatment (new page) groups

3️⃣ Statistical Testing (Z-test)

• Calculate pooled probability and standard error
• Compute Z-score for conversion rate difference
• Determine statistical significance (95% confidence interval)

4️⃣ Final Recommendation

• Launch new page if Z-score > 1.96
• Keep old page if difference is not significant

5️⃣ Visualization

• Bar chart showing control vs. treatment conversion rates
• Annotated chart with percentages for clarity

---

📈 Key Results (Sample Output)

• Control Group (Old Page) Conversion Rate: 12.0%
• Treatment Group (New Page) Conversion Rate: 12.3%
• Z-Score: 2.56
• Conclusion: New Page is statistically better. Launch it.

---

📊 Visualization

The chart visually compares conversion rates between control and treatment groups.

---

💡 Business Impact

✔ Quantified the impact of a new landing page
✔ Recommended data-driven decision on page launch
✔ Improved understanding of user behavior and conversion efficiency
✔ Visualized results for clear stakeholder communication

---

📌 Skills Demonstrated

• Python Data Manipulation (pandas)
• Statistical Analysis (Z-test)
• Data Cleaning & Deduplication
• Data Visualization (matplotlib)
• Business Decision Support based on Analytics

---

🚀 Next Steps (Optional but Impressive)

• Segment A/B test by device type or geography
• Run multi-page A/B experiments simultaneously
• Incorporate confidence intervals in charts
• Automate report generation for management dashboards

---

📦 Data Availability

The ab_data.csv dataset is included for running the analysis.
Can be replaced with any similar dataset for replication or further testing.

---

👤 Author

Y. Rithvesh
Python | Data Analytics | A/B Testing
Date: 15-02-2026