The new tool uses machine learning to match a shopper’s picture to similar-looking eBay products.

Online marketplace giant eBay Inc. will roll out an image search tool in its app and on the mobile web this fall.

Image search allows shoppers to take a picture, or select one saved on their smartphone, and enter it in eBay’s mobile search bar. That sends the marketplace on a quest to match the image to its 1.1 billion products and surface the most visually similar products on the search results page.

The new tool, announced Wednesday, is designed to help shoppers quickly find the products they want on eBay. When a consumer searches with an image, she doesn’t have to take the time to type words or think of the exact way to describe a product.

“With over 1.1 billion listings, it’s critical for eBay to organize the data of its vast inventory to make it easier to find and compare items,” says Mohan Patt, vice president of buyer experience at eBay.

The image search tool uses artificial intelligence and machine learning, which means that as more shoppers use it, the algorithm will learn and improve over time.



The retailer also announced a similar tool called Find It On eBay that allows consumers browsing on mobile devices to share a link of an image with eBay, which will then show similar products. For example, if a shopper is on Pinterest looking at coffee tables, she can tap the image, hit the share icon on her smartphone, select the eBay app, and the eBay app will launch and show similar-looking products.

“Moments of shopping inspiration can come at any time, whether you’re walking down the street or browsing your social media feed,” Patt says. “EBay Image Search and Find It On eBay make it possible for people to shop eBay using any image or photo that inspires them.”


Directly tying social networks to eBay’s e-commerce store is innovative and will likely lead to a better customer experience and thus sales, says Angela Culver, chief marketing officer of Akeneo, a product information management technology company.

“The ability to shortcut from an image directly to a product that’s available for purchase allows eBay’s users to connect in a powerful new way that brings products in the real world together with the capabilities of an e-commerce marketplace,” Culver says.

This new way to search with images dovetails with eBay’s larger structured data initiative, which also aims to better catalog products and help shoppers find more relevant search results by having sellers define their products with standardized terms. Structured data is the bedrock foundation for the image search tool, Patt tells Internet Retailer.

“By structuring eBay’s trove of data on the back end, machine-learning systems are able to layer on top and unlock artificial intelligence and computer vision algorithms to surface the most relevant items and improve discoverability for users,” Patt says.


This visual search tool likely will make eBay’s search results more accurate, says Josh Levine, co-founder and chief design officer at e-commerce design agency Cake & Arrow.

“The tool has the capacity to vastly improve search accuracy, leaving way less room for misinterpretation and for the loss of important details about products,” Levine says.

Although consumers may not use images as the primary way they search for products, it will be useful when shoppers want to find a product that is difficult to explain, such as a replacement part for an electronic device or a dress with a unique pattern, says Gareth Dismore, CEO at SearchSpring.

“EBay’s announcement is just the latest sign we’re seeing of leading retailers making it easier than ever for shoppers to find what they’re looking for,” Dismore says.


The marketplace will roll out image search on mobile first, then develop complementary technology to use on other platforms, like desktops, Patt says.

EBay built the feature in-house. The idea for Image Search came from its annual companywide competition called “eBay Hack Week” that challenges its technology team to develop new e-commerce ideas. Image Search won the competition in 2015, and eBay has worked on the feature every since, Patt says.

Other retailers, such as Neiman Marcus (No. 41 in the 2017 Internet Retailer Top 500) and Wayfair Inc. (No. 16), have launched similar visual image search tools on their e-commerce sites.

This is not the first artificial intelligence tool eBay has deployed. For example, eBay’s Shop The Look feature uses artificial intelligence to match different looks, such as striped outfits, to clothing pieces for sale on eBay. This tool is powered by technology that it acquired last year from visual search technology company Corrigon Ltd.


Also in 2016, eBay acquired two other vendors that specialize in machine learning and artificial intelligence: Expertmaker, based in Malmo, Sweden, which focuses on artificial intelligence, machine learning and big data analytics; and SalesPredict, an Israeli company that uses analytics to predict customer buying behavior and sales conversion.