Table of contents

If you talk to colleagues in the online marketing consulting field these days, you will inevitably hear the term A/B testing. True to the motto "test, test, test", so-called "spilt tests" try to get the most out of one's own content by testing different versions on the market and evaluating them for better results.

With this test method Two systems popularly used to test the original version of a piece of content against a slightly modified version and then evaluate it. Both versions are played out to an equal number of users and then evaluated on the basis of different performance indicators. But why do many agencies, vendors and brands bother?

On Amazon, this option was only available to merchants selling on for a long time. Since Q4 2021, these functions are also available on the European marketplace and are specifically available to German sellers. Sellers will be able to split product images and A+ content as well as titles, bullet points and product descriptions. There are two requirements that have to be met:

1. A Trademark registration must be set up for the respective product.

2. The desired product must be sufficiently Traffic have received. Only in this way are reliable analyses of the split tests is possible. ASINs with too little traffic may therefore not be displayed at all in the "Manage your experiments" function. For them, it is recommended to increase the traffic, for example, by SEO measures or PPC campaigns and then go into testing. Amazon does not reveal exactly how high the traffic has to be, and it is very likely that it is category-dependent.


Step 1

First of all, it should be determined, Which products or materials are to be tested for individual listings. Here, for example, products can be selected in descending order according to the biggest sales drivers in order to achieve the most noticeable effect possible. Also new products can be suitable for this purpose if enough traffic can be generated quickly.

When selecting the test, whether product images, bullet points A+ or similar, it all depends on which materials you want to test. We recommend a very clear sequence here: Main image / product gallery, SEO, A+ content.

Often, even very rudimentary subtleties such as a new quality seal, Made in Germany labeling or similar and can generate much more effective results than getting lost in optimized details. We must not forget the listing from the customer's point of view to consider: 

What does the customer see first? Usually the main image on the SERP.
Afterwards? The entire gallery of the product. 
What are they looking for in the process? According to USPs, seals, comparisons and product features... 

Here, not only can one's own Content strategy define, but directly in practice testto achieve maximum Performance out of the products. 🚀

Step 3

At Filling in the information In addition to mandatory information such as start date and test duration, there is also the option to store a hypothesis. With the Hypothesis there is the optional possibility to name the reason for the test, in order to better assign the results to the goals set up in advance.

Step 4

Test versions, too, must of course contain all Amazon guidelines fulfill. After sending the test variants, it should therefore be checked whether the Content approved has been performed. In addition, care should be taken to perform a maximum of one test per ASIN and to avoid multiple tests at the same time.

Step 5

At the Evaluation of the tests version A and version B should differ clearly from each other distinguishin order to be able to draw meaningful conclusions from the testing. In order to interpret the results of the Amazon A/B tests, the marketplace provides some KPI available. These include, for example: Conversion, units sold and the sample size. The probability that one version of the content is better than the other is also provided by Amazon. The result of an A/B test can also be that no clear results were obtained. This may be because version A and B were similar or the product page did not have enough traffic for meaningful results. However, it can also show that the tested product simply does not require any adjustments or that a different hypothesis should be tested.

In sum, the purpose of such tests is clear: You want to identify the best building blocks for your listing. With the help of A/B experiments, vendors and sellers can determine if Listing changes to a Higher conversion rate lead. You can run tests to gather insights into how customers respond to a version of the A+ content, title, bullet points, or main image. However, A/B tests should be evaluated with caution. The results are only statistically relevant if a high number of sales could be generated across both variants in the tests. With small numbers of cases, the probability is comparatively high that the Result only random has arisen.

If you want to sustainably improve the quality of your listing through A/B testing, please feel free to contact us. Contact with us. We offer you an individual analysis and thus provide you with targeted solution strategies.

Table of contents


NEWS 1st BIG PICTURE MEETING Dear friends of eFly, This month we had our Big Picture Meeting. The whole team was present and we took the opportunity to

Amazon Marketing News 34 - 51KW

1. last days of Christmas business This year's Christmas business is in its final stages. As of today (12/21), some items are still available until Christmas. Analogous to the last years products,