📊 +40% Conversion | 📈 Statistically Significant | 🎯 Targeted 18–25 Segment
📈 Credit Card Launch A/B Testing & Market Analysis 📝 Project Overview
This project explores whether a newly designed credit card should be launched to a wider audience. The analysis includes data validation, customer segmentation, A/B testing, and statistical hypothesis testing using a two-sample Z-test. The goal is to evaluate the performance of the new credit card compared to the existing one and provide a data-driven business recommendation.
🎯 Business Objective
Determine if the new credit card performs significantly better than the old one by comparing customer transaction behavior across test and control groups.
🚀 Phase A: Data Preparation & Market Understanding
- Data Validation
Validated raw CSV and Excel files received from a third-party data provider.
Performed sanity checks for missing values, duplicates, and structural inconsistencies.
- Data Import & Understanding
Loaded datasets into Python/Excel for exploration.
Reviewed data types, distributions, and basic summary statistics.
- Data Cleaning
Treated nulls, inconsistent formats, and outliers.
Ensured clean, standardized data ready for analysis.
- Exploratory Data Analysis (EDA)
Key insights:
18–25 age group shows highest potential for new card adoption.
26–48 age group requires competitive differentiation.
49–65 age group prefers low-maintenance, secure card options.
🧪 Phase B: Experiment Design & Campaign Execution
- Group Formation
Identified 246 customers aged 18–25.
Selected 100 customers for the test group (new credit card).
Created a 40-customer control group using the existing card.
- Campaign Performance
Campaign duration: 2 months
Conversion rate: 40% (40 out of 100 customers adopted the new card)
Daily comparison:
Control group performed better on 29% of days
Test group performed better on 71% of days
📊 Statistical Hypothesis Testing (Two-Sample Z-Test) Goal
Validate whether the improved performance of the test group is statistically significant.
Approach
Compared mean daily transactions between test and control groups.
Used Z-test (sample size > 30).
Calculated:
Z-score
Critical Z-value
p-value
Results
Z-score > critical value
p-value < 0.05 (alpha)
✔ Rejected the null hypothesis
Conclusion: The new credit card significantly outperforms the existing card.
🟦 Final Recommendation
The new credit card shows strong performance and statistically validated success. It is recommended for full-scale market launch, starting with the 18–25 demographic.
🛠 Tools & Technologies
Python: Pandas, NumPy, Matplotlib, SciPy
SQL: Data extraction & cleaning
Excel: Validation & preprocessing
Visualization: Jupyter Notebook (Matplotlib, Seaborn)
Statistics: A/B Testing, Z-test, Hypothesis Testing
💡 Skills Demonstrated
Data validation & cleaning
Customer segmentation
Exploratory data analysis (EDA)
A/B testing & experiment design
Hypothesis testing (Two-sample Z-test)
Statistical interpretation
Business insight generation
End-to-end analytics workflow
📬 Contact
Naima Tanveer Email: naimatanveer49@gmail.com
LinkedIn: www.linkedin.com/in/naimatanveer