Can Artificial Intelligence Sell Shoes?
Source: ELIZABETH DWOSKIN
Searching for the right item in an online store can be a numbing experience, an endless parade of product images that crowd screen after screen. Now a footwear retailer believes it has a solution: Artificial intelligence.
Shoes.com is teaming up with AI specialist Sentient Technologies Holdings Ltd. to offer a novel way to search for shoes on its Canadian domain www.shoeme.ca. (Its U.S. site will continue without AI for now.) Starting Tuesday, the site will offer an “visual filter” option in addition to the standard filtered search. The software will try to learn users’ preferences by examining their behavior.
The AI-powered visual filter offers an array of images of shoes, inviting users to click on one that is most like what they are searching for. From there, the software continues to suggest new options as it refines its understanding of the user’s taste. Clicking on a purple pair of galoshes, for example, yields more purple galoshes, as well as galoshes in other colors, and galoshes in different patterns and designs. Clicking on designs brings up more designs. Clicking on shoes with laces yields options with laces. And so on.
Artificial intelligence has become a hot technology in Silicon Valley, especially software that can find subtle patterns in clicks and other online behaviors and use them to serve up what users want before they ask for it.
PJ Worsfold, Chief Innovation Officer of Shoes.com, said that he hoped the new option would help streamline the shopping experience by helping people find what they wanted more quickly. Sometimes, while shopping online, “you kind of hit a wall and you are mired in all this choice,” he acknowledged.
He expected that the new software to expand users’ tastes. “Because most [product recommendation] algorithms recommend what others bought before, the result is that you are going with the crowd.” With AI, he said, he expects to see “a wider distribution” of style preferences.
Mr. Worsfold said he planned to learn whether the visual discovery method yields more clicks, more interactions, and ultimately more purchases than the current features.
The immediate benefit of the visual filter wasn’t entirely clear because shoes.com already offers many filtering options such as size, width, brand, price, and type, enabling a consumer to use language to refine their searches fairly thoroughly.
Some hope that new methods will help ecommerce capture a large share of the retail pie. While ecommerce jumped 14.1% in the second quarter of 2015 compared with the same period of 2014, according to the U.S. Census Bureau’s most recent quarterly update released in August, it accounts for just 7.2% of overall retail sales. And growth is starting to flatten. New third-quarter estimates of U.S. ecommerce activity will be released on Tuesday.
Visual search has been a longstanding challenge for ecommerce companies, said Sucharita Mulpuru, a retail analyst at
Forrester Research . Amazon.com announced in 2010 a visual search feature and an “Items That Look Like This” feature on product details pages. A press release at the time said Amazon’s software analyzed product images for shape, color, appearance, and classifies similar-looking products together. Today on Amazon, other products are recommended through a tab that says “Customers Who Viewed This Item Also Viewed.”
Ms. Mulpuru said Sentient’s technology was impressive, but no company had “cracked the nut.”
Technology giants and startups say they are making progress in using AI to help people sift through the tide of visual images on the Web. Alphabet Inc., the parent company of Google, and Yahoo Inc. have launched tools that attempt to pick out objects in photos, though the technology is far from perfect. Pinterest recently launched a tool to help users find and shop for items on the company’s site. EyeEm, a photography marketplace, uses a visual discovery engine to help publishers find photographs that will be most appealing to viewers.
Sentient is considered a heavyweight in this area because its founder, Antoine Blondeau, helped developed the technology behind Siri, Apple, Inc.’s artificially intelligent personal assistant. The startup has raised over $100 million and uses AI for financial trading. Blondeau has said he hoped to expand into health care, medical research, fraud detection, public safety, and other areas ― evidence of the rising aspirations of many computer scientists specializing in AI.
Ms. Mulpuru said that better ways to sift through a dizzying amount of content online could help boost overall ecommerce sales, but the biggest impediment to online apparel sales was that most shoppers prefer to try things on.
“Even if you have look-alike products,” she said, “you still don’t know what they look like on you. That’s the bigger obstacle.”
| }
|