E-Commerce App User Behaviour

Project Title: E-Commerce App User Behaviour Analysis

Link notion page: (English version) (versi Bahasa Indonesia)

Dashboard Preview:

Description: This case study analyzes user behavior and conversion performance of an e-commerce mobile application using event-level data. The analysis focuses on understanding the user journey from product views to purchases, identifying key funnel bottlenecks, engagement patterns, and revenue-driving products and categories. By combining session-level analytics, funnel analysis, and product performance insights, this study provides actionable recommendations to improve conversion rates, user engagement, and overall business performance.

Dataset: https://www.kaggle.com/datasets/mkechinov/ecommerce-events-history-in-electronics-store

Analytical Approach

Methods

The analysis was conducted using an event-based analytics approach combined with session-level aggregation to understand user behavior across the e-commerce funnel. Key methods include:

• Data Cleaning & Validation: Ensuring data quality through checks on missing values, duplicates, timestamp ranges, event types, and price validity.
• Sessionization: Aggregating event-level data into session-level metrics to analyze user journeys and conversion behavior.
• Funnel Analysis: Measuring user drop-off across key stages (View, Cart, Purchase) to identify conversion bottlenecks.
• User Behavior Analysis: Analyzing engagement metrics such as events per session, session duration, and high-intent sessions.
• Product & Category Performance Analysis: Evaluating revenue, conversion, and cart abandonment at product, brand, and category levels.

Dashboard Structure

1. Summary Overview
   • High-level KPIs: DAU, MAU, revenue, session conversion rate, average session metrics
   • Revenue trends and top categories/brands
2. Funnel & Conversion
   • Funnel visualization (View, Cart, Purchase)
   • Conversion rates, cart abandonment rate, and session-level conversion metrics
   • Category-level conversion comparison
3. User Behaviour
   • User engagement metrics (events per session, sessions per user)
   • DAU vs MAU comparison
   • User activity heatmap (hour & weekday)
   • High-intent sessions analysis
4. Product & Category Performance
   • Top categories and products by revenue
   • Cart abandonment by product/category
   • Price vs conversion relationship
   • Category, Brand, Revenue matrix

Tools

• PostgreSQL Data ingestion, cleaning, transformation, and creation of fact and dimension tables
• Power BI Data modeling, DAX measures, and interactive dashboard visualization

Key Insight

• The biggest conversion bottleneck occurs at the View-to-Cart stage, with only 8.46% of sessions progressing to Cart.
• User activity peaks during morning to early afternoon hours (approximately 07:00–14:00), while evening activity is lower than expected.
• Users who reach the Cart stage demonstrate high purchase intent, with a Cart-to-Purchase conversion rate of 57.66%.
• A significant number of high-intent sessions (many views and cart actions) fail to convert into purchases.
• Revenue is highly concentrated in computer and electronics categories, with the top 3–5 categories contributing the majority of total revenue.

Business Recommendations

• Optimize Product Discovery & Listing Pages Improve product visuals, descriptions, pricing visibility, and recommendation logic to increase View-to-Cart conversion.
• Leverage High Purchase Intent at Cart Stage Maintain and further optimize checkout experience, payment options, and cart UX to preserve the strong Cart-to-Purchase conversion rate.
• Reduce Revenue Dependency on Top Categories Explore cross-selling strategies and promotional campaigns for mid-tier categories to diversify revenue sources.
• Optimize Campaign Timing Schedule push notifications, promotions, and marketing campaigns during morning and midday peak activity periods.
• Investigate High-Intent Non-Converting Sessions Conduct deeper analysis on pricing competitiveness, stock availability, trust signals, and payment friction to reduce lost conversions.

The PBIX file is too large to upload directly to GitHub. Please download it from the following link: https://drive.google.com/drive/folders/1B4PeHTzAdUC63dDFpdXwEcBh7n7XSisH?usp=sharing