Split test
Main image optimization and A/B testing
Situation
The product had an existing main image that showed the megaphone from a limited angle. The image did not communicate the complete product package or highlight its important features effectively. Customers had limited visual information to understand the product and its key selling points at first glance.
Task
My responsibility was to improve the product's visual presentation and make the listing more informative. The goal was to create a stronger first impression and improve the chances of converting visitors into customers.
Action
I started by analyzing competitor listings to understand how similar products were presented on Amazon. Then I studied their main images and gallery images to identify common visual approaches and opportunities to differentiate our product.
I then used DataDive to conduct keyword research, identified relevant search terms and studied their relevance to the product. This helped me understand what customers were looking for and which product features deserved greater attention in the listing.
Based on my research, I redesigned the main image to present the product from a clearer angle; improved its visual presence, focused on communicating important product features and included components more effectively.
Instead of relying only on my design judgment, I tested the new creative against the existing version through Amazon Seller Central. I monitored the experiment to evaluate its impact on conversion rate, units sold and total sales.
Result
The new creative performed better than the original version across all the major reported metrics.
The conversion rate increased from 0.486% to 0.672%. This represented a 38.3% relative improvement.
Total units sold increased from 255 to 374. This was an increase of 119 units.
Total sales increased from $28,664 to $42,211. This represented a difference of $13,547 during the experiment.
Units sold through search also increased from 147 to 225.
Amazon's experiment reported an 87% probability that Version B was better than Version A. Based on the experiment results, Version B was selected and automatically published.
This project demonstrated how competitor research, keyword analysis and data driven creative testing can work together to improve Amazon listing performance.
Some more results
Here is a glimpse at some more optimizations and their respective split test results.
Dibyamurty Shaw · 2026
Made in Framer