In a rare fireside chat at SXSW, Amazon Fashion chief technology officer Tony Bacos mapped out how the company’s using machine-learning in its bid to crack online retail’s biggest challenges.

According to Bacos, the spate of initiatives spans virtual fitting, personalized recommendations and computer vision: “The opportunity to create better experiences online, both from a discovery-and-browse capability, the fit challenge and being able to, I like to say, search with your eyes, not your keyboard, all of those are challenges that may have been solved in little bits and pieces. But we’re trying to tackle all of them.”

It’s a far cry from how critics typically think of Amazon: When it comes to fashion, the marketplace is known as a place to pick up basics. And it works best when customers know what they’re looking for. Think search, not discovery.

Amazon has an advantage, though: It’s a tech giant and a major developer of artificial intelligence with deep pockets. In the holiday quarter, the company posted $3 billion in net income, a jump of 66 percent over the previous year. Because of its massive consumer electronics, cloud and e-commerce operations, it can afford to splurge on efforts to boost its fashion business.

Top of the list are “personalization and helping customers sort through a growing catalog of products,” Bacos said. Amazon is using AI and machine-learning to help “connect customers with the products we know are going to delight them, that fit their tastes, their style, as well as their bodies.”

Amazon Fassxswhion

Fashion influencer Blair Eadie on stage with Amazon Fashion’s Tony Bacos.  Adriana Lee

Fit is a tough challenge, even for Amazon. Brands have different sizing charts, and people of the same size can prefer looser or tighter clothing. These are all issues that Prime Wardrobe was created to address. The try-before-you-buy service allows shoppers to try on items at home before purchasing. It also happens to add to Amazon’s wealth of customer data.

“Being able to have that particular customer’s preference, as well as their body shape, their purchase history — whatever they’re willing to share with us, so that we can give them a more confident fit experience — is a big area of investment for us,” said Bacos.

Another is visual search, so people don’t have to labor over the right key words or descriptions, and what the company internally refers to as its “outfit builder.” The feature leans on computer vision and machine-learning to recommend products that complement a particular item.

Bacos, who previously worked on the Alexa team, confirmed Amazon’s interest in bringing its voice platform and fashion together more. As it is, the connection exists primarily in the form of the Echo Look fashion selfie camera.

“Part of the challenge with voice-only experiences for highly visual categories is bridging the gap and helping customers understand what they can expect,” he said. “What we see right now is that a lot of customers are interested in talking with Alexa about fashion, about clothing, and there are things that customers are saying with some frequency — things like, ‘Alexa, what should I wear?’…we also, over time, are going to build more and more multi-Modal experiences that are both voice and screen.”

He described a scenario in which shoppers start with a voice interaction and continue on the Alexa app or Fire TV. Amazon’s popular line of streaming video devices has had Alexa support since 2017. The same year, the company also debuted display-equipped Alexa devices.

This complex matrix of tech-fueled approaches seems rather ironic, considering the somewhat simplistic motivation for Amazon’s fashion fixation to begin with.

“For us, this is a really exciting category because it is popular,” Bacos added. “Whether you call it ‘fashion’ or you just call it ‘pants,’ [it] has some amount of appeal to everybody.…And nobody has really nailed it yet.”

load comments
blog comments powered by Disqus