Amazon's new algorithm COSMO, collectible

Recently, Amazon officially launched its new algorithm COSMO, which is quite different from the A9 algorithm and may subvert the existing keyword ranking mechanism. This article is divided into 6 parts, explaining the Amazon COSMO algorithm and sellers' response plans in detail.

What is COSMO?

COSMO (Customer-Oriented Search & Match Optimization) is an advanced technology developed by Amazon .

Simply put, COSMO is a super brain of Amazon. It generates useful common sense knowledge by analyzing and learning the purchasing behavior, search habits, etc. of a large number of users, and better understands the intentions of buyers.

What does COSMO mean to buyers?

For buyers, the system analyzes their past shopping behaviors and search habits, interprets their search behaviors, understands the ultimate intention of the search, and recommends products that they may be interested in, thereby improving shopping efficiency and satisfaction .

Learn about the COSMO algorithm

COSMO can accurately push products based on user information, shopping history and other big data to match user purchasing intentions , rather than just matching through search keywords.

For example: A pregnant woman wants to buy a pair of shoes and searches for "maternity shoes" in the Amazon search box. The COSMO algorithm will analyze the "importance of anti-slip and other functions to pregnant women" based on the large amount of "human common sense" stored in large language models such as GPT or OPT and ask "the reason why the user buys or co-purchases", and then recommend anti-slip shoes and other products that can capture the pain points of pregnant women's needs to increase conversion rate.

Another core function of COSMO is multi-round navigation, which is to further explore consumers' potential needs through intelligent recommendations.

For example, a search for “camping” may lead to the option of “air mattress” and then be refined to “air mattress for camping.” COSMO will then provide various types of air mattresses for camping to meet different needs, such as lakeside camping, mountain camping, or 4-person camping.

This feature is Rufus, which was exposed some time ago. It helps customers search, discover, research and purchase products in a "question and answer" way.

Rufus has entrances in the top search box and product details page of the Amazon app.

For example, when I enter the details page of a dress, I am too lazy to read the details page, so I can directly ask if this dress is suitable for a 1.8-meter-tall person. Rufus will answer and recommend based on buyer comments, QA, listing and other information.

Comparison of COSMO and A9 features

Summary of the key points of COSMO's new model

1. Recommendations that are more in line with people’s search intent: Understand users’ search intent and provide recommendations that are more in line with user intent.

2. Make recommendations by analyzing known order information: Use historical order data to improve the accuracy of recommendations.

3. Filter products that are too similar: reduce duplicate recommendations and improve user experience.

4. Improve the accuracy of common-sense recommendations for complementary and alternative products through intelligent algorithms: optimize recommendation algorithms and improve recommendation quality.

5. The original page conversion rate data is still very important in recommendation: use historical conversion rate data to improve recommendation results.

6. Filter by specific attributes and adjust search results based on personal preferences: Provide personalized search results to meet the needs of different users.

• What sellers need to do is to study buyers carefully and understand their preferences and needs. Optimize product detail pages and highlight product highlights to make your products easier for buyers to discover.

• What sellers need to do is not just focus on keyword rankings, but also pay attention to product quality and cost performance, provide high-quality after-sales service, and improve buyer satisfaction.

• Buyers are tired of the same old products. Now they prefer unique and creative products.

• Sellers need to innovate products, develop new features, create differentiated products, pay attention to market trends, and seize emerging demands.

• Amazon’s advertising mechanism will not change, and the role of advertising will remain important.

• What sellers need to do is to learn how to place advertisements, improve the advertising effect, control the advertising costs and obtain the maximum benefits.

• What sellers need to do is to carefully fill in product tags, including category tags, selling point tags, attribute tags, etc. Optimize product titles and descriptions to increase product appeal.

• What sellers need to do is to analyze the target customer groups, understand their shopping habits and behavioral characteristics, and use data analysis tools to optimize store operation strategies.

With the internal testing of Amazon's COSMO system, our operating strategies as sellers are bound to change. In the A9 algorithm era, keyword optimization is king, but in the COSMO era, user intent and purchasing behavior will become the focus.