Amazon E-Commerce Sales & Customer Behavior Analysis

Data Analyst Portfolio Project — MySQL

This project analyzes 1,000,000 Amazon e-commerce transactions to identify actionable insights across sales performance, product/category economics, customer behavior, seller quality, pricing/discount strategy, returns, delivery operations, and inventory risk.

Project Snapshot

• Transactions: 1,000,000
• Customers: 603,815
• Products: 89,999
• Sellers: 9,000
• Categories: 5
• Subcategories: 16
• Brands: 12
• Cities: 5
• Analysis period: 2024-03-31 to 2026-03-31

Tools & SQL Skills

• MySQL
• Common Table Expressions (CTEs)
• Window functions: LAG(), DENSE_RANK()
• CASE WHEN bucketing
• GROUP BY and aggregations
• HAVING
• Date-based analysis
• Return-rate analysis
• Seller quality analysis
• Business-focused KPI analysis

Key Business Findings

1. Electronics generated 66.3% of revenue while representing about 17.7% of orders, driven primarily by much higher average order value.
2. Monthly realized revenue remained broadly stable around the ₹33.5–₹38.0 Cr range in the report period.
3. Overall return rate was 11.60%; the 5–7 day shipping group had a 14.39% return rate.
4. The report identified a significant seller-quality gap: high-revenue sellers were not necessarily highly rated.
5. Customers averaged about 1.66 orders per customer, indicating a broad-reach, relatively low-repeat marketplace.
6. The 40%+ discount band had more orders but substantially lower revenue than the 10–19% band.
7. 39,659 products were below the 20-unit stock threshold.

This repo contains the SQL query file, the full PDF report, and this README.

Important Note

The PDF contains selected SQL examples rather than the complete 30+ query source set. The SQL file in this repository therefore contains the query logic visible in the report. If you have the original full .sql query set, replace/add it before presenting the repository as the complete 30+ query project.

The report also documents a data-quality issue involving delivery_status and is_returned. Revenue queries use is_returned = FALSE, while delivery_status is used for delivery-performance analysis.

Portfolio Report

See the PDF report in the report/ folder for the complete business analysis, outputs, insights, and recommendations.

Author

Mujahid Khan
Data Analyst

• LinkedIn: https://www.linkedin.com/in/mujahidkhan-data-analyst/
• GitHub: https://github.com/mujahidk902