Optimizing Checkout Upsells: A/B Testing for Product Prompts in E-Commerce
Company
BundleJoy
Goal
This project applies A/B testing methodologies to evaluate the effectiveness of such prompts in an online shopping environment. By splitting users into two groups—one that sees the upsell prompt ("Show Prompt" group) and one that does not ("No Prompt" group)—we can measure the impact of this intervention on key e-commerce metrics.
Tools used
-
Programming Language: Python
-
Data Analysis: Pandas, NumPy
-
Visualization: Tableau, Matplotlib, Seaborn
Dashboard
This dashboard presents key insights from an A/B test evaluating the impact of displaying a product upsell prompt during checkout.
All metrics are designed to help quantify the business impact of the upsell prompt and guide data-driven decisions on checkout optimizations.
Key Analysis and Takeaways
Users shown the upsell prompt were significantly more likely to add the product to their order.
95% Confidence Interval for uplift:
($0.60, $14.86)
Bayesian analysis reveals a 98.25% probability that the upsell prompt increases total order value.
Estimated total revenue gained from the prompt:
$7,577.46
Average revenue uplift per prompted user:
$7.74
No negative impact on conversion or increase in cart abandonment.
Visualizations
Queries
Total revenue per group (A/B Test)
SELECT group,
COUNT(*) AS total_sessions,
SUM(total_order_value) AS total_revenue,
AVG(total_order_value) AS avg_order_value
FROM checkout_sessions
GROUP BY group;