What is Amazon AB Testing?
We spoke with Andrew Browne, Co-Founder of Splitly, an Amazon AB testing service that helps sellers identify successful product content and pricing to increase conversions.
According to Splitly, most products are not optimized for maximum sales. For example, how do sellers know if their pricing is correct or which image performs best? The bottom line is unless you test, you just don’t know.
“Sellers should have a fully optimized listing on Amazon because there is a direct correlation between a good listing and increased profits,” Browne said.
We spoke with Browne on the value of good content and how Amazon AB testing contributes to the overall success of a seller’s strategy on the Marketplace.
Q. Why is Amazon content a critical component in the Marketplace?
The first aspect to good content is about being identified by Amazon’s algorithm for any relevant search queries.
You want to make sure that your product listing is appearing for anyone that would be remotely interested in your product.
How sellers appear in the search results is influenced by which keywords they include in their:
- Product Title
- Key Product Features
- Product Description
- Search Terms
The second aspect of having a good Amazon listing is about actually converting visitors into customers once they land on your page.
This is where it is important to optimize for your best product images, marketing copy, and price.
It is important to note that this mainly applies to private label sellers, who have full control over their product listing. If you are one of many sellers on a given listing, and competing for the Buy Box, then you may not have the ability to edit the product listing, in which case these optimizations are a moot point.
Q. What are some of the major challenges or pitfalls Amazon sellers face with content optimization?
Challenge #1: Setting Up Proper Tests
Amazon AB testing is not like split testing on a regular website. With regular web traffic, you can alternate which variation a user sees on every single visit if you wanted. However, Amazon will only provide traffic data for an entire day.
This means that in order to run a proper split test on Amazon, you must change the listing every single day at midnight Pacific (which is when Amazon starts a new day). And you will have to do this for at least two weeks or more, depending on the amount of traffic your listing gets. This is what you need to do in order to maintain data integrity.
So, how does Amazon AB testing impact sellers?
Splitly is the first automated split testing tool, created specifically for Amazon sellers.
After you link your Amazon accounts to Splitly’s secure platform, you set your goal (whether conversion rate, profit, clicks, etc), your desired statistical significance, and we handle the rest, so that you can run perfect split tests and easily identify listing optimizations.
The benefits of these split test results is twofold:
Firstly of course, you generate more profits with your optimized listing.
Your tests will identify which price point, keywords, images, and more will yield the greatest profits.
Secondly, Amazon rewards you for improved conversion rate by moving you up in organic rankings.
This is because Amazon’s algorithm identifies that your products convert at a higher rate than competitors, and when you make money, Amazon makes money. So there is a virtuous cycle of improved conversion rate and sales, and improved organic ranking for main keywords.
Challenge #2: Gathering the Data
If you have more than one ASIN you are selling on Amazon, gathering the split test data can be incredibly tedious. You need to record all of the relevant data (sessions, conversions, conversion rate, sales, etc) for each ASIN, one day at a time.
This is a very time-consuming challenge, more than what most people want to do.
Challenge #3: Interpreting the Data
Another big challenge people face when split testing on Amazon is interpreting the data correctly. It is hard to resist the human psychology of wanting to find tangible results, and jumping to early conclusions.
For example, a seller may set up a test, and in the earliest stages see a difference in performance between a control and variation. Within a few days, the seller may then determine that one variation will perform better than the other, but in reality this is pure luck because the tests have not been running long enough.
People trust their intuition more than actual data. But the challenge is really just having enough patience to accumulate enough data!
Q. What tips can you provide for improving Amazon product content?
Keyword Optimization: I recommend doing keyword research to see what phrases and keywords people search for the most. Use Google Keyword Planner to get ideas and a rough idea of how many searches per a month (this is based on Google data, not Amazon, but is helpful to assess relative search volume).
Put your main keywords towards the front of the title, and don’t repeat keywords. You can also add keywords in the Main Product Features (the bullet points) and the Product Description as well. You don’t need to repeat any words, as it doesn’t have any benefit to your keyword ranking.
Use the Search Term fields for any leftover keywords related to your product.
Don’t hold back here, Amazon now allows 1000 characters for each of the 5 fields.
I also recommend focusing on getting the best product images as possible.
And once you have a library of good images, test different images as your Main Image to see which generates the most clicks. I have seen people do some incredible things with 3D modeling images, which is good for some types of products. You can find people who do such things on Upwork or Fiverr.
For more on Amazon AB testing, email [email protected]