eBay introduced the first conversational AI shopping agent in May 2025 to a limited number of platform users in the US, and in October the same year, the company’s CEO announced that Mercury, eBay’s internal Agentic AI platform, is fully operational.
The early results show that Mercury is already affecting the visibility of product listings on the platform, as the click-through rate in organic search results declined by 70% when the AI overview appears at the top of search results for a specific query.
Consequently, sellers who exclusively optimize their listings for eBay’s Best Match algorithm and use conventional SEO strategies risk losing traffic, a decline in brand visibility, and potentially lower sales.
Still, this doesn’t mean that Generative Engine Optimization has entirely replaced product listing optimization. Rather, sellers must incorporate AI visibility strategies into their SEO approach on eBay marketplaces.
Read on to find out more about agentic search on eBay and the best GEO practices that will help you get your products in front of high-intent buyers.
Key takeaways
- eBay’s conversational shopping agent makes it easier for shoppers to discover product listings that are hard to find.
- Mercury prioritizes structured metadata, detailed information about a product, and content quality over keywords.
- 20% to 25% of product listings lack sufficient information to be included in AI-generated search results.
- Achieving and maintaining Top Rated Seller status serves as a confidence signal to eBay’s Mercury.
- GEO on eBay increases a listing’s chances of being cited in Google’s AI Overviews by 80% due to Mercury’s platform native integration with Google’s Vertex AI.
Understanding how eBay’s agentic search works
Unlike other major online marketplaces, eBay has dedicated significant resources to developing an internal infrastructure that powers its recommendation and search ecosystem.
Hence, Mercury isn’t only a shopping assistant but a full-fledged AI search engine that manages a chain of agents that sits atop the Krylov AI platform and works hand in hand with eBay’s Cassini search algorithm.
Mercury features a Listing Matching Engine capable of processing Large Language Models-generated text and mapping it across eBay’s inventory of over 2 billion items, restructuring the data in the process to make it searchable.
In other words, whenever a buyer enters a specific prompt, Mercury analyses it and breaks it down into entities and constraints.
So, for example, if a user is looking for a pair of black running sneakers, the platform treats ‘sneakers’ as an entity, while ‘black’ and ‘running’ serve as constraints.
Afterward, it commences the so-called vector search or the semantic mapping process, during which the platform looks for ‘the area’ in the database containing product listings with matching data, pictures, and seller reputation fitting the buyer’s search criteria, delivers value, and shows them to the buyer.
It’s also important to note that Mercury is iterative, so if the first set of products the platform recommends doesn’t meet the shopper’s requirements, the agent will ask a follow-up question to ensure the user has found items that match their search intent.
GEO as an extension of product listing SEO
Even though Mercury fundamentally changed how shoppers discover products on eBay, it didn’t render the conventional search engine optimization practices obsolete.
Hence, conducting keyword research and front-loading keywords in product listing titles is as important as ever. The only difference is that, unlike with traditional search, Mercury uses keywords as coordinates that help it connect a listing to the correct category.
So instead of searching for an exact keyword match, Mercury uses keywords as a starting point to track listings with high trust scores that match the buyer’s intent.
Still, repeating the same keyword too often or using its synonyms extensively can send a spam signal to the model and cause the listing to be excluded from the list of items the agent shows to the buyer.
In addition, all traditional SEO practices, such as ensuring each listing includes high-quality product images and relevant information about the item’s features and condition, remain essential for listing performance in search rankings and inclusion in AI recommendations.
Generative Engine Optimization only evolves conventional listing optimization by emphasizing populating the Item Specific section in each listing and structuring product descriptions to ensure they are readable by LLMs.
Most importantly, eBay’s agentic search platform cross-references text and images in listings, and if it detects any discrepancies, it reduces the listing’s confidence score.
That’s why visual and textual information included in a listing must align in order for the agent to label it as trustworthy and mention it in AI responses.
The handshake between eBay’s internal agentic search and Google’s AI Overviews
Recent data suggests that the vast majority, over 86%, of product listings that appear in Google’s AI Overviews for ecommerce-related searches come from eBay.
The reason for eBay’s dominance in AI overviews isn’t just its AI tools that automatically optimize listings for generative AI systems.
The online marketplace’s agentic AI engine is built on Google Cloud’s Vertex AI and Vector Search technologies, as eBay uses them to run A/B tests and handle semantic queries.
Consequently, Mercury seamlessly integrates with Google’s agentic search engine through the Universal Commerce Protocol (UCP), which is why it treats eBay product listings as credible sources.
So, whenever Google’s Search Augmented Generative Engine (SAGE) uses Gemini 3 Pro to look for products across the web to recommend in the AI Overviews shopping results, it prioritizes high-signal data.
eBay listings are compatible with Google’s AI crawlers because Mercury uses Vertex AI embeddings, enabling Google’s agent to verify products on eBay with high confidence scores.
Consequently, optimizing their listings for eBay’s internal generative engine helps sellers improve visibility of their products and attract traffic from multiple channels by getting brand mentions featured in AI Overviews.
Top GEO for eBay listings practices
As eBay continues to innovate and improve its agentic AI platform, sellers must adapt to these changes to ensure their products are discoverable by shoppers.
Standard product listing optimization practices are no longer as effective as they once were, which is why merchants must master GEO to stay competitive across eBay marketplaces.
We’ve shortlisted some of the best GEO practices for product listings on eBay to help sellers increase the visibility of their products, so let’s take a look.
Verify all information you include in a listing
Data quality is one of the primary trust signals for AI models, as it allows them to accurately allocate a product listing to an appropriate conceptual map.
As a result, populating the Item Specific section provides a model with a wealth of information it can rely on when determining which product listings match the buyer’s query.
Hence, verifying whether the product’s name, size, color, condition, product identifiers, and all other data are accurate improves the confidence score eBay’s agentic AI assigns to a listing and boosts its search visibility.
Additionally, using bulleted lists to structure the text and highlight an item’s key features makes it more readable for generative engines. In turn, it increases a listing’s chances of being included in the products the AI agent recommends to a shopper.
Invest in the production of high-quality pictures and video content
The quality of a listing’s visual content was an important ranking factor before Mercury was launched, as eBay’s Cassini search algorithm was measuring the CTR that listing’s images were getting to determine its position in the search results.
Unlike Cassini, the Mercury platform has the ability to analyze the content of product images and videos and cross-reference it with information included in the listing’s title, product description, and Item Specifics sections.
So, sellers should update their content strategy and follow specific image and video guidelines, such as ensuring their product images are 1600×1600 pixels and that the product occupies approximately 85% of the frame, to create GEO-friendly content.
Additionally, sellers should use 10 to 15-second full-HD vertical videos that show products from all sides to ensure that the Mercury platform can use their video content in its multimodal reasoning.
Write listing titles and product descriptions in natural language
Mercury processes semantics to understand the meaning of a listing’s title or its product description.
That’s why titles and descriptions written in natural language allow Mercury’s NLP model to understand the relationship between each word and assign the appropriate confidence score to the listing.
Additionally, writing naturally helps the conversational AI to answer the questions shoppers ask during their buying journeys. That’s why mentioning the best ways to use a product or how to upkeep it can be an effective GEO strategy that boosts a listing’s visibility.
It’s also worth noting that Mercury penalizes keyword stuffing by performing keyword tokenization to prevent merchants from confusing its vector map and tricking it into placing a listing into the wrong category.
Amplify confidence signals by maintaining the Top Rated Seller Ranking
Seller Performance metrics are an important ranking factor that both eBay’s internal search engines use to rank and recommend products to shoppers.
Consequently, sellers with a Below Standard ranking are likely to be pushed down in the search rankings.
On the other hand, Mercury uses Seller Performance metrics as signs, so merchants with high Transaction Defect or Late Shipment rates are unlikely to have their products featured in the results shoppers see when using eBay’s conversational AI.
The platform evaluates the sellers’ confidence score in real time, so a high number of Item Not As Described claims in a single week can affect a merchant’s confidence score even if their Seller Performance rating for the month is Above Standard or Top Rated.
That’s why maintaining the Top Rated Seller ranking and building trust with consumers are essential GEO strategies that help sellers reach buyers through both conventional search and AI-generated product recommendations.
Expand your inventory to get more organic traffic
The size of an eBay store’s inventory can be an important factor in helping Mercury determine whether a seller is an authority in a given category.
That’s why gradually expanding a specialized inventory can be a reliable GEO strategy, as such inventories serve as knowledge hubs that enable the LLMs to understand a seller’s position within different product categories and cross-reference an item’s price, size, and other features.
So, adding new products to an eBay store regularly can help merchants maintain a high confidence score, since AI models often view sellers who don’t update their product portfolio for a long time as unreliable.
It’s also worth adding that eBay’s agentic AI prefers new listings and often favors them over older listings with little or no activity or marginal purchase history over longer periods, which is why frequently adding new products to their inventory can help sellers secure a steady source of organic traffic.
Master Generative Engine Optimization to get your products in front of the right buyers
Organic traffic is still the most reliable path to profitability on eBay. Still, the platform’s search ecosystem is evolving as Generative Engine Optimization is becoming increasingly important, while getting a product listing onto the first page of search results is becoming less efficient.
eBay businesses must adapt to these changes by incorporating GEO into their SEO strategies and ensuring their products are featured in AI-generated responses or sources cited that shoppers see when using the marketplace’s conversational AI.
Optimizing each listing for eBay’s Cassini and Mercury algorithms doesn’t have to be a time-consuming task. Schedule a call to learn how Webinterpret can help automate your listing optimization process and build a strong brand presence across eBay marketplaces.
About Webinterpret
Webinterpret supports merchants selling on eBay.
Our AI-based solutions enable more effective selling through automated listing localisation, optimisation, advertising, and returns, and ensure all products placed on EU markets are GPSR-compliant.
By giving your international and domestic customers a full, end-to-end local shopping experience, Webinterpret improves your conversion rates and helps establish your business locally and globally.
