E-commerce Customer and Sales Analysis
Project Overview
This project was developed as part of the Data Analyst Internship Program at Mayerfeld Consulting.
The objective was to analyse an e-commerce dataset containing customer, transaction, website session and marketing campaign information. The analysis focused on understanding the factors influencing revenue, refunds, customer behaviour and conversion.
Business Questions
The project explored questions such as:
- Which product categories generate the most revenue?
- What factors are associated with website conversion?
- Are refunds concentrated around particular products or reasons?
- How concentrated is revenue among the company’s customers?
- Do website behaviour metrics relate to purchase value?
- How effective are different marketing campaigns and acquisition channels?
Approach
I began by exploring the structure and quality of the datasets, identifying missing values, inconsistencies and the relationships between the different tables.
After preparing the data, I used Python to perform exploratory data analysis and statistical hypothesis testing. This allowed me to distinguish between patterns that appeared visually interesting and relationships that were statistically supported.
I then created an interactive Tableau dashboard presenting the most relevant results through revenue trends, geographical comparisons, category performance, refund analysis and customer behaviour metrics.
Key Insights
Electronics generated the highest overall revenue. This also contributed to a high financial value of refunds within the category, although no single refund reason clearly explained the result.
Converted website sessions had a higher average time on site than non-converted sessions, suggesting that stronger engagement may be associated with conversion.
The relationship between pages viewed and purchase value was statistically significant but practically negligible. This highlighted the importance of evaluating the magnitude of a relationship rather than relying only on its p-value.
Revenue was not strongly dependent on a very small group of customers. The ten highest-spending customers represented only a small proportion of total net revenue, indicating a relatively diversified customer base.
Outcome
The project strengthened my ability to manage an end-to-end analytical workflow, from data preparation and statistical analysis to dashboard development and insight communication.
It also reinforced the importance of connecting every visualisation and statistical result to a clear business question.
Tools
Python · pandas · Statistical Hypothesis Testing · Tableau · Data Visualisation · Data Storytelling