17/01/2023
*When Split Testing on Amazon is Beneficial*
Split testing compares two different versions of an element to see which one performs best
It is an effective way to optimize elements of a product listing
To accurately measure each change, we split test only one element at a time
We use controlled experiments between two different versions of an element based on specific metrics
🔴 Low sessions and/or CTR - customers are not engaging enough to click on a listing
𝘴𝘰𝘭𝘶𝘵𝘪𝘰𝘯: split test with improvements to pricing, main image and title keywords
🔴 Low conversion rate - customers are clicking but not converting.
𝘴𝘰𝘭𝘶𝘵𝘪𝘰𝘯: split test with different secondary images, bullets points and product descriptions
We run an initial split test, off Amazon, using the services of PickFu, to receive real test data from real people
We confirm these results by running a live split test on Amazon using Manage Experiments to test simultaneously
At the conclusion of the test, we use the results interpreted by Amazon to determine which variation is likely to be more profitable
In the example below, title variation B had a better conversion rate than A, producing more sales from the same sample size
By monitoring and recording significant changes to sessions, CTR and conversion rate we get an indication of the impact of the creative testing
If we see an increase, we know the split testing was effective and will result in an increase in sales
If we see a decrease in these metrics and a decline in sales, we accept that the testing results may have been inaccurate and we will run a new test