Can Gen-AI shopping agents improve the shortcomings of algorithm-based search engines? If the answer is yes, an improved shopping experience could give eBay sellers who are frustrated with the current state of Search reason to celebrate. But as sellers have learned, execution is everything – and change can be very, very disruptive.
The expansion to “agentic AI” may be eBay’s biggest move since it changed the default sort order of its search engine to “Best Match” 17 years ago. Former eBay manager Adam Nash explained at the time – not long after leaving the company in 2007 – that until Best Match, eBay search had been extremely “literal” and “transparent”:
“Until changes were made in the last few years, eBay search would literally do only the following:
1) Look at the keywords entered by the buyer
2) Look at the title keywords of every listing on the site
3) Return only the listings that had 100% of the keywords entered by the buyer
4) Sort the listings by “time remaining””
Since the introduction of Best Match, eBay looks at a multitude of factors when deciding how to rank listings in search results, such as price, shipping costs, delivery speed, seller feedback ratings, and buyer behavioral patterns. While eBay views those abilities as positive, its ability to “pull levers” along with a lessening of transparency generally hasn’t been viewed positively by sellers.
Nash called eBay’s 2008 move to make Best Match the default in search “bigger than anything I can think of in the history of the eBay buyer experience.” Agentic shopping has the potential to do that again.
When eBay announced on Tuesday it had begun rolling out its new AI shopping agent to some shoppers, it said it would show up wherever the user was in their buying journey, either by reacting to a user’s request or through predictive messaging inline on the page the user was visiting.
“Seamlessly woven into your shopping journey, this intelligent agent delivers real-time, hyper-personalized product picks and expert guidance based on your shopping preferences as you explore our marketplace. From pinpointing the perfect gift for your best friend to assembling the ideal Spring Break wardrobe, our agent makes discovery effortless, fun, and uniquely tailored to you.”
Presumably eBay’s AI shopping agent is powered by the OpenAI Operator agent, which eBay had disclosed in January it was using in order to “help buyers discover more of the things they love, and enabling sellers to be successful and grow their sales.”
In its May 6th announcement, eBay said its new AI shopping agent was an evolution “from simple chatbots to conversational and interactive agents that are equipped to understand, interact, and anticipate user needs,” adding, “Such capabilities have the potential to enable us to lead the next generation of shopping and selling, fueling marketplace growth, and solidifying our position in AI-powered ecommerce.”
Rivals are not ignoring agentic shopping either. Etsy is using “Algotorial curation” and posted the following video on YouTube last month:
And big-dog Amazon launched Rufus in February of 2024, calling it a “a generative AI-powered expert shopping assistant trained on Amazon’s extensive product catalog, customer reviews, community Q&As, and information from across the web to answer customer questions on a variety of shopping needs and products, provide comparisons, and make recommendations based on conversational context.”
eBay said its new AI shopping agent is currently available to a small percentage of US customers and will become available to more users on a rolling basis.
Some important and unanswered questions remain for sellers. How might the AI shopping agent impact Promoted Listings that sellers use to boost visibility on eBay?
And, as was the case with Best Match, agentic agents may further decrease transparency into how eBay selects listings to display to shoppers using the tool. That may be among the biggest concern for sellers.
Update 5/11/2025: Updated for clarity.
It would be hard to imagine a more chaotic situation than unleashing the AI agent against buyers looking for high quality stamps with a specific Scott catalog number. Ebay does not currently support catalog numbers or item condition in the Stamps category, two of the most important filters for stamps. Categories and crude number matching allow potential buyers to specify catalog numbers in search. Most searches return a manageable number of extraneous result but searching for catalog numbers like 1 or 500 is hopeless. Buyers may try searches like “US 1” or “US 500” hoping to find some hits but some listings will be missed. Buyers in the Stamps category hate to miss listings in their price or quality range because every stamp is different.
Will an AI-driven “best match” search dredge up more grossly overpriced listings or misidentified stamps? Sponsored ads are not encouraging. Yesterday when I was searching for some specific Monaco stamps for pricing, four sponsored listings exposed four US stamps. Two of the stamps were misidentified as substantially more valuable stamps. This may explain grumbling about eBay on philatelic chat board and be the reason that some prominent philatelists are afraid to buy on eBay.
How do eBay BOTS assigned to “read” reports about questionable listings determine if items are genuine and not misidentified or counterfeit?
As a former stamp collector I must agree that stamps would be among the worst things for A.I.
There are variations in color, which is made worse by the fact that different electronic devices will initially capture those colors differently and display those colors differently. The old tube monitors for instance display black better than newer flat screens. These devices can be set to capture and / or display images using different color profiles.
Different physical sizes; reissues of the same series; perforation sizes, different designs; centering; missing perforations.
Watermarks that can’t be seen with the naked eye.
Repaired stamps in which missing areas are filled with grafted paper – you need to examine under a microscope.
A cancellation, such as a rare Dead Post Office (DPO) can make a item more valuable. Stamps still on covers are typically worth more.
Fakes, cinderella material, reprints, altered stamps, fake cancellations, fake covers.
Very complicated – I’ve only mentioned a few issues.
Hopefully, it won’t let all of the Chinese vendor magically float to the top.