What Actually Drives E-Commerce Sales: Reviews Beat Ratings, Discounts Backfire

By shrijeetverma13 · August 14, 2026

This analysis of 50,000+ products reveals surprising patterns in what makes items sell. While most retailers assume star ratings matter, the data shows…

Review count has a strong positive correlation (0.901) with units sold, meaning products with more reviews tend to sell significantly more units. Rating, on the other hand, shows only a weak correlation (0.188) with units sold, suggesting a product's star rating alone doesn't strongly predict how many units it sells.

Winter is the top-performing season by revenue, generating $118.5 million from 1.8 million units sold, while Summer actually sells more units overall (nearly 2 million). Among styles, Casual is the clear winner, leading both in revenue ($97 million) and units sold (2.24 million), making it the strongest performing style by far. Two bar charts break down revenue and units sold by season and by style, and supporting data tables show the full breakdown of these figures.

Discounts have very little effect on units sold and revenue overall. The correlation between discount percent and units sold is nearly zero (0.045), and the correlation with revenue is slightly negative (-0.013). While the highest average units sold (280.9 units) occurs at the steepest discount level (41-50%), the highest total revenue ($268.8 million) actually comes from products with no discount at all. This suggests that while big discounts can increase how many units you sell, they cut into revenue enough that selling at full price generates more total money.

Dresses is the top-performing category, generating $100,076,862 in total revenue from price times units sold. Two charts were created to visualize this: one showing total revenue across all categories, and another highlighting the top 10 subcategories by revenue, giving you a clear picture of where sales dollars are concentrated.

Dresses have the highest return rate among categories (13.9%), costing an estimated $14.1 million in lost revenue. Denim is the top material for returns (13.0% return rate, $10 million lost), and Bohemian is the leading style (11.4% return rate, $6.4 million lost). These segments represent the biggest opportunities to improve profitability by addressing likely issues with sizing, quality, or product descriptions.

Analysis identified 3,200 products (6.4% of the catalog) with high stock levels but low sales, indicating overstock risk. Products were flagged when stock quantity was in the top 25% (53+ units) while units sold were in the bottom 25% (21 or fewer units). A scatter chart highlights these risky products against the rest of the catalog, and a bar chart ranks the top 15 products by their stock-to-sales risk ratio, helping pinpoint which items need attention first.

The analysis grouped sales data by category and color, calculating average price, total units sold, and total revenue for each combination. A bar chart and data tables have been generated showing the top 15 category-color pairs ranked by total revenue, highlighting which combinations of product type and color drive the most sales value.

A correlation matrix was calculated for price usd, discount percent, rating, review count, stock quantity, units sold, and return rate, and visualized as a heatmap. The heatmap uses a red-blue color scale to show how strongly each pair of variables moves together, with values ranging from -1 (strong negative) to +1 (strong positive), making it easy to spot patterns at a glance.

Using the IQR method on price usd across the whole dataset, 1,412 products were flagged as outliers, with the normal price range falling between roughly -$32.82 and $147.74 (any price above $147.74 counts as a high-price outlier). Box plots were generated to visualize the price distribution and outliers by category, giving you a clear picture of how prices spread out and where the extreme values sit across different product categories.

Materials were ranked by both average return rate and average rating to spot mismatches between customer satisfaction and product returns. Denim has the highest return rate (12.98%) but also ranks 4th in rating, while Linen ranks 2nd highest in returns (10.91%) and 3rd in rating — both flagged as potential quality-perception mismatches since they're well-liked yet frequently returned. On the other end, Leather has the highest average rating (3.127) and a relatively low return rate, making it the most consistent performer. A combined table, a scatter plot comparing rating vs return rate, and a bar chart comparing the two rankings side-by-side were all generated to visualize these patterns.