Changes and updates to Amazon's algorithms are closely related to every Amazon seller, as these changes can alter how products are displayed, affect search rankings, and ultimately influence sales outcomes. The recent announcement of Amazon's new COSMO algorithm has caught the attention of Amazon sellers.
The implementation of the new algorithm indicates that Amazon's traffic distribution mechanism might experience a significant adjustment. Will the existing product ranking advantages of sellers continue to exist, and how will the inherent search ranking logic change? Is the new algorithm a disruptive change to the original A9 algorithm or an advancement of it, and how should Amazon sellers think about their future operational direction?
1. What is the Amazon COSMO Algorithm?
Amazon's COSMO (Customer-Oriented Search & Match Optimization) algorithm is a new artificial intelligence algorithm developed based on large language models (LLMs). Its core lies in analyzing user behavior data to uncover users' potential shopping intentions and constructing a customer-centric knowledge graph.
Compared to the previously used A9 algorithm, the COSMO algorithm places greater emphasis on individual user needs and the purchasing experience. While the A9 algorithm mainly determined search result rankings based on keyword matching, sales, and reviews, the COSMO algorithm uses big data and artificial intelligence technology to delve deeper into the common-sense knowledge hidden behind user behavior, building a comprehensive and accurate industry knowledge graph.
The COSMO algorithm, by mimicking the human brain, better understands what users are looking for during searches. It looks beyond the keywords entered by the user and guesses what the user might want to buy, thereby providing more personalized product recommendations.
2. Is Amazon's A9 Algorithm Going to Become a Thing of the Past?
Based on the information about the new algorithm that has been released, the COSMO algorithm refers to a machine learning algorithm that optimizes product search rankings and recommendation systems with a user-centric approach. This algorithm focuses on mining deeper shopping intentions from user behavior by analyzing data such as users' shopping history and browsing habits, intelligently pushing products that the user may be interested in, thereby achieving a more personalized shopping experience.
The COSMO algorithm is somewhat different from the well-known A9 algorithm. The A9 algorithm determines the appearance and position of products in buyers' search results based on relevance and performance factors such as sales ranking, price, conversion rate, click-through rate, reviews, and buyer satisfaction. The core of the COSMO algorithm, on the other hand, is about mapping users and products into a common vector space. By calculating the similarity between user vectors and product vectors, it predicts products that users may be interested in and determines the priority of recommendations accordingly.
Does the introduction of the new algorithm mean that A9 will become a thing of the past?
Actually, that's not the case. As the "housekeeper" of the Amazon platform, the A9 algorithm, which Amazon prides itself on, is bound to continue to exist. A9, as the core of Amazon's SEO search engine, will still play an important role in the future search mechanism.
The introduction of COSMO is more of a complement and upgrade to the A9 algorithm, especially in terms of understanding user intent and enhancing the precision of the search and recommendation system. COSMO enhances buyer user experience and search efficiency through more precise identification of user intent and personalized recommendations. Therefore, the A9 algorithm will not become history but will continue to evolve to adapt to new trends in the e-commerce field.
3.What changes has the COSMO algorithm brought to Amazon's internal site?
The COSMO algorithm's paper mentions that COSMO not only extends the knowledge graph to 18 of Amazon's main categories but also that COSMO has been successfully deployed in various Amazon search applications, including search relevance, recommendations based on user dialogues, and search navigation. It also indicates significant progress has been made.
Some veteran Amazon sellers have also felt the fluctuations of the new algorithm. Some sellers noticed that the rankings of some newly introduced niche products are steadily climbing, while some mainstream products that once dominated the top of the charts seem to have lost their competitive edge. Sales and rankings no longer hold their high positions, and even price promotions are not as effective as before.
This indicates that due to the introduction of the new algorithm, the results buyers get from keyword searches will likely no longer be those uniformly best-selling products. Instead, on top of A9's base, there's an additional calculation of the buyer's potential intent, allowing deeper buyer needs to be met and enabling more precise searches for products that satisfy those needs.
Take maternity shoes as an example. The traditional A9 algorithm might only recommend ordinary "maternity shoes" to buyers based on relevance and performance, without considering the special needs of pregnant women for anti-slip features.
However, the COSMO algorithm, based on the vast "human common sense" stored in the AI large language model and inquiring about "the reasons for user purchases or joint purchases," analyzes the "importance of anti-slip and other functions for pregnant women," thus recommending products like anti-slip shoes that better meet the actual needs of pregnant women.
This recommendation mechanism based on the AI large language model not only helps to increase user satisfaction and improve the shopping experience but also can promote the growth of sales. The recommendation engine contributes more than 35% to the revenue of the Amazon e-commerce platform.
The COSMO algorithm also improves search navigation by adding multi-round navigation to help users precisely express and meet their purchase intentions. For example, when searching for "camping" products, the algorithm will refine recommendations based on the user's continuous choices, presenting options such as "air mattresses," "tents," "blankets," "lanterns," and other camping products. If a buyer selects "air cushion," it further refines to "camping air cushion," greatly enhancing the accuracy and user experience of the search.
4. How should sellers adjust their operational direction with the launch of the COSMO algorithm?
These changes suggest that the focus of sellers' operations should no longer be solely on "keyword indexing," but rather on making more optimizations around the product itself, improving aspects such as appearance, functionality, color, etc., to create product differentiation. Sellers should diversify and comprehensively enhance the precision of their listings, perfecting all the details within the link to enable more accurate AI-driven recommendations.
- What exactly do users want?
If previously the advantage for sellers was to grow in scale and strength, now it's about being more clever and refined. Gaining a deeper understanding of the user's profile and preferences is essential for achieving higher exposure.
- Refine the attributes, tags, and keywords of the product
Sellers need to rethink how they organize their product listings to align closer with buyers' purchase intentions and product characteristics. At the same time, they should gain a deeper understanding of their customer base, grasp the shopping habits and preferences of the target consumers, build more detailed audience profiles, and enrich the tags in their listings with more finely categorized attributes.
The more detailed the product categorization, the more likely it is to be recommended. For example, with light foods, it could be segmented into: light foods for women, postpartum light foods, suitable for summer, convenient for work takeout, suitable for belly slimming, dinner-specific, etc.
- Innovation and differentiation to avoid product homogeneity
In terms of products, focus on more niche offerings to avoid homogeneous competition and change the mindset of price wars by developing more products that meet individual buyer needs.
- Adjust marketing strategies flexibly
Based on market changes and algorithm updates, promptly adjust marketing strategies, including pricing, promotions, and advertising, to maintain competitiveness.
