Amazon’s new algorithm is born, will the cross-border circle be “reshuffled”?

Recently, Amazon's new COSMO algorithm has become a hot topic in the cross-border circle because an Amazon team published a related report in Amazon Science. So the circle is saying that the A9 algorithm will become a thing of the past, and a new algorithm is coming. In the future, Amazon will usher in earth-shaking changes. Some sellers even attributed the recent order cut to this new algorithm.

What is COSMO? What is the difference between it and the A9 algorithm? What changes will the application of this algorithm bring to future operations? How should sellers adapt to this new algorithm? Is the A9 algorithm going to be phased out? ... Sellers urgently need to know the answers to these questions.

Amazon tests new algorithm in the US

In fact, before making COSMO public, Amazon had tested it rigorously and the results were good.

Earlier, Amazon conducted an A/B test on 10% of the users in the US site for several months and found that after integrating COSMO into its online search navigation system, the sales of products in this market segment increased by 0.7%, equivalent to a surge of hundreds of millions of dollars in annual revenue. In addition, it was observed that in the same traffic market segment, the navigation participation rate increased by 8%, which shows that customer interaction and satisfaction have been greatly improved.

Based on the success of the initial test, Amazon expects that by expanding the adaptability of COSMO to cover all navigation paths, it will have the potential to generate billions of dollars in revenue growth. More importantly, for Amazon, COSMO's huge potential in various other functions and applications opens up new ways to enhance user experience and business growth.

Having said so much, what exactly is COSMO?

According to the report, COSMO stands for "Amazon Large-Scale E-Commerce Knowledge Generation and Service System".

"Common sense" means that you need to wear clothes when it's cold and you can't wear cotton clothes when swimming. Amazon has built such a common sense system under the e-commerce model. The report lists an example: if a user searches for "pregnancy shoes", according to common sense, pregnant women should not fall, and shoes need to have "anti-slip" function, so the shoes presented to the user are those with relevant functions.

This is similar to the targeted recommendations of domestic e-commerce companies that are tailored to each individual, that is, the system feels that consumers may need it more and it may be more suitable.

Amazon believes that understanding the user intentions behind a large number of noisy behaviors in online e-commerce platforms will be of great benefit to many downstream applications, such as recommendation and product search. From the perspective of cognitive science, intention is the psychological state of human beings committing to take action, and behavior is the result of intention.

For example, "In order to attend a wedding banquet, we need to buy ordinary clothes." The intent of "attending a wedding banquet" is used to rationalize and explain the user behavior of "buying ordinary clothes." In online shopping scenarios, if e-commerce platforms can accurately capture user behavior, they can provide users with explainable recommendations and personalized search experiences in a smarter and more friendly way.

However, this intent is not something that people explicitly express. It requires common sense to understand, and therefore is challenging for machines to extract in a scalable way.

COSMO therefore leverages the power of large language models (LLMs) to mine user-centric common-sense knowledge from a large number of behaviors and build a broad knowledge graph.

It improves search relevance by injecting commonsense knowledge into the search process, making search results more aligned with user intent.

Simply put, the COSMO algorithm uses AI to "artificially think" about user searches and guess what products users want to buy.

Although Amazon has only introduced 10% of the market share, based on the current test results, it is only a matter of time before it is widely introduced. However, for now, sellers still have many questions to be answered.

The fate of a group of sellers will be reshaped

As we all know, A9 is the algorithm that Amazon has long used for search and advertising ranking, which mainly sorts search results based on keyword matching and relevance.

According to the report, the COSMO algorithm pays more attention to understanding user intentions and purchasing behaviors. It predicts user purchase intentions by analyzing big data such as user purchase history, browsing behavior, and personal preferences, thereby providing more personalized search results and recommendations.

Obviously, the new algorithm has overturned the keyword ranking mechanism. You should know that the A9 algorithm competes on the sales volume under the same keyword, which is why there is the current spiral strategy, and it also leads to the fact that the top products under the same keyword are almost all the same products.

Under the new algorithm, Amazon sellers’ product sales have become diversified, and they may get more orders even if their keyword rankings are not high. Therefore, many people believe that the new algorithm can effectively alleviate the current situation of low-price circulation, which is a good thing for small and medium-sized sellers and can greatly increase their chances of survival.

Not only that, after the birth of the COSMO algorithm, differentiated products will have more opportunities, and competition among highly homogenized products will become increasingly fierce because the new algorithm will automatically filter out similar products.

It can be predicted that under the COSMO algorithm, the fate of a large number of sellers will be reshaped, so many industry insiders say that the cross-border circle will usher in a major reshuffle. Many sellers are very anxious about this.

In fact, sellers do not need to worry too much. First of all, no matter how things change, they still remain essentially the same. For sellers, products are the core. Without good products as a foundation, no matter how many external operations are done, they are in vain. Therefore, even if COSMO is widely used, product development and iterative updates are still the main tasks of sellers. On this basis, consider how to obtain this new traffic entrance by optimizing titles and other operations.

Secondly, some sellers said that the platform will still be based on keyword/search e-commerce in the future, that is, the A9 algorithm will not be abandoned, so the operating methods under this algorithm can still be used. However, if sellers want to open new traffic entrances under the new COSMO algorithm, they need to do some research on the personalized needs of buyers.

So what should Amazon sellers do to take advantage of COSMO’s features? Possible strategies include:

Enhance product listings: Sellers must optimize product titles, descriptions, and metadata to make them as descriptive and clear as possible, with a focus on aligning with customer intent and common-sense associations.

Leverage data insights: Sellers should use the data provided by Amazon to understand how customers discover and interact with their products. This information can guide adjustments to product offerings and marketing strategies.

Stay aligned with customer intent: As COSMO grows, it will continue to discover new customer trends and preferences. Sellers need to be flexible and adjust their strategies based on these insights.

Encourage customer engagement: Post-purchase feedback and reviews are crucial. Not only do they contribute to the knowledge graph, influencing future search relevance and recommendations, but they also provide valuable insights for sellers to improve their products and customer service.

COSMO Algorithm