As we all know, if we want to get continuous orders on Amazon as sellers, we must improve the ranking of our products to the first page of search results as much as possible. The higher the ranking, the more traffic we get; the more traffic, the higher the sales.

Amazon has its own algorithm for determining which products appear at the top of search results. Experienced sellers are familiar with its name, the A9 algorithm. The algorithm is trained to prioritize product listings with a good sales and conversion history, rather than simply keyword-filled product listings.

Similar to other search engines, Amazon’s A9 algorithm takes into account a variety of factors to rank products.

Algorithm changes lead to abnormal increase in new product rankings?

Before I share with you the new Amazon algorithm, I would like to share with you an analysis chart of the A9 algorithm by a foreign research institute:

All experienced sellers have their own understanding and cognition of Amazon's A9 algorithm ranking, and formulate corresponding strategies based on these cognitions to improve the ranking of products.

However, recently some sellers have found that their category rankings have changed significantly compared to last year:

In this way, user search results will no longer be the same products, and the user experience will obviously be better.

According to the previous ranking logic, this situation should not occur in theory. Sellers speculate that Amazon may have changed its algorithm.

You in front of the screen, I wonder if the above changes have occurred in your product category? Sellers who want to discuss can add Zhaodanmao WeChat: FLA66668888, and ask customer service to add you to the group for discussion.

A US seller also noticed the changes in Amazon's ranking mechanism: "In March of this year, the Amazon team extracted 10% of the keyword search results to test the new AI ranking mechanism COSMO, which was a huge success and increased the conversion rate by 0.7%. It is estimated that this part of the optimization can bring an additional sales of 4.9 billion US dollars a year."

According to him, COSMO can comprehensively consider big data such as user information and shopping history to match users' purchasing intentions, rather than just relying on the matching of search keywords. Please refer to the figure below for details.

What exactly is COSMO?

Zhaodanmao found relevant reports on the official website of Amazon Science. The official explanation is: Although the existing e-commerce knowledge graph integrates a large number of concepts or product attributes, it is unable to discover user intentions and there is a gap between people’s thinking, behavior, and the way they interact with the world around them.

To this end, Amazon proposed COSMO, a scalable system that can mine user-centric common-sense knowledge from massive behaviors and build industry-scale knowledge graphs to enable diverse online services.

Simply put, COSMO is a large-scale e-commerce common sense knowledge generation and service system developed by Amazon. It aims to mine user-centric common sense knowledge by analyzing a large number of user behaviors and build an industry-scale knowledge graph. Amazon has expanded COSMO to 18 major categories of Amazon and deployed it in various Amazon search applications, including search relevance, session-based recommendations, and search navigation. The characteristics of the COSMO system are as follows:

User intent understanding: COSMO aims to mine users’ implicit intent by analyzing their behavior and use this information to improve the intelligence of Amazon’s various online services.

Common sense knowledge: It builds a knowledge graph by extracting knowledge from large-scale language models and training with manually annotated data, so that the service can better understand and respond to users' actual needs.

Application scenarios: COSMO's knowledge graph can be used to improve search relevance, the accuracy of recommendation systems, and improve search navigation, etc.

If Amazon subsequently widely applies the COSMO system to platform algorithms, it will undoubtedly subvert the existing ranking mechanism.

Under the previous A9 algorithm, our sellers focused on keywords; under the COSMO system, sellers need to adapt to the new AI ranking mechanism:

With the internal testing of Amazon's COSMO system, the operation strategy of sellers will inevitably change. In the A9 algorithm era, keyword optimization is crucial, but in the COSMO era, user intent and purchasing behavior will become the focus.

In the future, Amazon’s search rankings may more accurately reflect consumers’ purchasing intentions, which will be beneficial for improving user experience and conversion rates.

Therefore, as sellers, if we foresee changes in things in advance, we must continue to learn and adapt to these changes, deploy in advance, ensure that our product links conform to the search mechanism of the new algorithm, and obtain better ranking positions and traffic!

Sellers who want to study this algorithm in depth can take a look at the full English paper written by the Amazon team on COSMO!