E-Commerce Sales Analysis

Overview

What the data shows

This analysis examines 2023 transaction data from a Turkish online retailer to understand what drives revenue and purchasing behavior. The results show that revenue is concentrated in a small number of product categories and driven primarily by customer retention rather than discounting, device type, or delivery speed.

Long-term growth should prioritize repeat purchasing behavior over short-term tactical optimizations like blanket discounting or delivery speed improvements.

Analysis

Five business questions

Each question below is answered with a chart and a written finding.

Click to expand.

1 Which product categories drive the most revenue vs. order volume?

Revenue in 2023 was highly concentrated among a small number of categories. Electronics, Home & Garden, and Sports accounted for 81% of total revenue despite representing only 38% of units shipped.

In contrast, Beauty, Food, and Books shipped nearly identical unit volumes but together contributed just 7% of total revenue. The data reveals a clear split between high-value categories that drive revenue efficiency and lower-value categories that rely on volume without a proportional return.

Pie chart showing total revenue by product category — Percent Revenue by Product Category (2023) — click to enlarge

Bar chart showing total revenue and order volume by product category — Total Revenue and Volume by Product Category (2023) — click to enlarge

2 Do higher discounts actually lead to higher revenue per order?

Discounted orders had slightly higher pre-discount order values than non-discounted orders. On average, orders receiving discounts were approximately $39 higher before the discount was applied. However, this does not establish that higher discount amounts drive increased order revenue.

Bar chart of AOV by discount status — Average Pre-Discount Order Value by Discount Status (2023) — click to enlarge

3 How do returning customers differ from new customers in spend and frequency?

Returning customers differ from single-purchase customers primarily in purchasing frequency, not spend per order. The average order value between the two groups differs by only $27, while returning customers place 6.2× more orders.

This indicates that revenue growth is driven by customer retention rather than higher per-order spend. Strategies such as loyalty programs or re-engagement campaigns that increase purchase frequency are likely to outperform those focused on upselling.

Bar charts comparing average order value and order volume between new and returning customers — Order Value and Volume by Customer Type (2023) — click to enlarge

4 Does device type affect purchasing behavior and order value?

Device type shows a clear relationship with purchasing behavior but only a modest impact on order value. Mobile devices account for the majority of orders, while Desktop orders have an average order value approximately $61 higher than Mobile orders. Tablet usage remains relatively low across both order volume and value.

Overall, these results suggest that device choice primarily influences purchasing frequency rather than spend per transaction, with Mobile devices serving as the dominant channel for engagement and Desktop devices providing a small advantage in per-order value.

Bar chart showing AOV by device type and pie chart of total orders by device type — AOV and Percentage of Orders by Device Type (2023) — click to enlarge

5 Is faster delivery associated with higher customer ratings?

Faster delivery is not strongly associated with higher customer satisfaction. Average ratings remain nearly flat across all delivery time ranges, with only a slight decline as delivery time increases.

Given the minimal variation, investments aimed specifically at reducing delivery times are unlikely to yield meaningful improvements in customer ratings.

Bar chart showing average customer star rating by delivery time range — Average Star Rating by Delivery Time Range (2023) — click to enlarge

Workflow

Excel vs. Python

After completing this analysis using Excel Pivot Tables, I reproduced and validated those results in Python using Pandas and Matplotlib. As expected, results were identical. Thus, the differences that matter are in workflow.

For a clean, moderately-sized dataset like this one, Excel was the practical choice: faster to set up, easier to present, and no environment to configure. Python's overhead wasn't justified at this scale, but the tradeoffs reverse quickly as data volume, reproducibility requirements, or team size grow.

Dimension	Excel (Pivot Tables)	Python (Pandas + Matplotlib)
Ad-hoc speed	Fast setup, no environment needed. Excel wins	Less efficient for one-off exploration on small, clean datasets.
Reproducibility	Steps are manual and less auditable. Risky for anything repeated or handed off.	Methodology is documented in code by default. Python wins
Scalability	More than adequate for ~14K rows. Best for this specific use case	Offers superior scalability and integrates naturally into automated pipelines.
Visualization	Built-in charts are fast and presentation-ready for standard visuals. Best for this specific use case	Simple charts require significant code, but Matplotlib offers complete control.
Collaboration	Familiar to most stakeholders and easy to share without technical setup. Excel wins	Better for technical teams requiring version control via Git.

↑ Back to Top