This project analyzes user behavior in an e-commerce platform to understand how users move through the conversion funnel:

View → Cart → Purchase

The objective is to identify drop-off points, evaluate user intent, and provide data-driven recommendations to improve conversion rates.

This project combines SQL (core analysis) with Python (EDA & visualization) to deliver a complete analytical workflow.

• The heatmap visualizes user activity across different hours and days
• Darker regions indicate higher engagement levels
• Activity is concentrated during afternoon hours and weekends

• High engagement periods do not always align with high conversion periods
• This reinforces the need to separate activity vs intent behavior

• End-to-end funnel visualization (View → Cart → Purchase)
• Drop-off analysis across funnel stages
• Conversion performance by price segment
• Time-based behavior (hourly & weekday patterns)
• Key insights panel for quick decision-making

The dashboard is designed to provide a clear business narrative, not just metrics, enabling stakeholders to quickly identify optimization opportunities

• Source: E-commerce Events Dataset (November 2019)
• Records: ~67 million events
• Type: Event-level user interaction data

• event_time
• event_type (view, cart, purchase)
• user_id
• product_id
• price
• user_session

data/ → raw dataset
notebooks/ → Python EDA
sql/ → structured analytical queries
images/ → visual outputs
dashboard/ → (optional future work)

1. Data validation and quality checks (SQL)
2. Funnel construction (user-level)
3. Sequence-based funnel refinement (true user journey)
4. Price-based segmentation analysis
5. Time-based behavioral analysis
6. Business insights and recommendations

• View Users: 3.69M

• Cart Users: 826K

• Purchase Users: 441K

• View → Cart: 22.36%

• Cart → Purchase: 53.45%

• Overall Conversion: 11.95%

• View Users: 3.69M

• View → Cart Users: 823K

• Purchase Users: 361K

• View → Cart: 22.28%

• Cart → Purchase: 43.86%

• Overall Conversion: 9.77%

👉 Sequence-based funnel reveals ~80K false conversions in the loose model, providing a more accurate representation of user behavior.

• Low-price users show high browsing but low intent
• Medium-price users demonstrate the highest conversion
• High-price users also convert well, indicating intent-driven behavior
• Conversion is influenced more by intent than price alone

⚠️ Note: Presence of zero-value price records may slightly affect segmentation accuracy

• Peak activity: 2 PM – 5 PM
• Highest conversion: 9 AM – 11 AM
• Weekend activity highest (Fri–Sat)
• Best conversion day: Sunday (~2.2%)

• Users browse in the afternoon but convert in the morning
• Friday shows high activity but lower conversion → browsing-heavy behavior

• Largest drop-off occurs at View → Cart stage (~75%)
• Only ~9.7% users complete the funnel in correct sequence
• Loose funnel overestimates conversion → sequence matters
• Medium & high-value users show stronger purchase intent
• User behavior varies significantly across time and intent levels
• High-activity users contribute a disproportionately large share of purchases, indicating that engagement depth is a key driver of revenue

• Better product recommendations
• Clear pricing & value communication
• UX optimization

• Focus on medium & high-value segments
• Personalize offers based on behavior

• Run campaigns in morning hours
• Push promotions on Sundays

• Simplify cart & checkout experience
• Build trust signals (reviews, guarantees)

• Improving the View → Cart conversion rate by just 5 percentage points could result in approximately +180K to +200K additional purchases

• This highlights that small improvements in early funnel stages can lead to significant downstream revenue impact due to scale

• The View → Cart stage represents the highest leverage point for optimizing overall business performance

• High-activity users show significantly higher conversion rates, indicating stronger purchase intent and engagement

• Low-activity users are primarily browsing users with weak intent, contributing heavily to early-stage drop-offs

• This suggests that user engagement depth is a strong predictor of conversion likelihood

Targeting high-activity users with personalized offers can significantly improve conversion efficiency

• SQL (PostgreSQL) → core analysis (CTEs, aggregations, funnel logic)
• Python (Pandas, NumPy) → data preprocessing
• Matplotlib / Seaborn → visualization
• Jupyter Notebook
• Power BI → interactive dashboard development

This analysis highlights that while user engagement is high, conversion is limited by weak intent formation in early funnel stages.

By aligning product strategy, pricing, and marketing efforts with user behavior patterns, businesses can significantly improve conversion performance.

• Build interactive dashboard (Power BI / Tableau)
• Apply cohort analysis
• Add user retention tracking
• Explore A/B testing strategies