In a year that knee-capped apparel and designer retail, launching a fashion shopping app might seem like madness. But for The Yes cofounders Julie Bornstein and Amit Aggarwal, the risks were worth it — and not just for them, but for the fashion world at large.
The way Bornstein explained it, when The Yes debuted in May, it was on something of a retail rescue mission.
“Our heads were all spinning from what this coronavirus thing was in late March, which is when our original launch date was going to be,” she told WWD. “But by the time May came around, we were like, ‘Listen, we want to do everything we can to help drive volume for these brands.’
“Because a lot of their traditional channels were not open: their stores weren’t open, their orders were being canceled,” she continued. “We see ourselves as a partner in all senses of the word, a technology partner that can help.”
The desire to help brands through the pandemic has become a common refrain from retail tech platforms. But few know the challenges as innately as Bornstein, having years of experience at companies like Nordstrom, Sephora and Stitch Fix.
With $30 million in funding, thanks to a Series A in October 2019, she and ex-Googler Aggarwal had the resources and tech know-how to bring something different to app-based fashion e-commerce.
Spring 2021 Fashion Collections for Women: Joy
From the consumer side, The Yes looks deceptively simple. Users look at various tops, dresses, shoes and other products, and easily like or dislike them — sort of like Tinder, but for shopping instead of dating, and quick “yes” or “no” taps replacing swipes.
The Yes uses a yes/no model to learn shoppers’ tastes and style preferences. Courtesy image
Of course, there’s much more going on below the surface. Those interactions yield data that helps zero in on the customer’s preferences, refining the selection of looks that land in front of a given user. It’s the sort of prediction modeling that was already considered the holy grail for retail before the pandemic, but could be a critical lifeline for brands now.
Either way, the momentum seems undeniable, especially for The Yes.
Since its launch in May, the company reports as many as 5.5 million yes/no interactions on the platform and a partnership roster that’s grown to 225 brands, from Balenciaga and Prada to Everlane, Frame and Levi’s. Meanwhile, the app has ballooned to cover nearly 69,000 stockkeeping units.
The Yes didn’t disclose hard sales numbers, but did say sales have doubled over the holiday season so far — which is notable, because the app isn’t really geared for gift-giving, nor does it focus on promotions or discounts. In fact, two-thirds of its sales come from full-priced items.
As for loyalty, 36 percent of shoppers come back within a month to buy again. And in a recent survey asking users if they’d prefer shopping in the app over places, the company said 90 percent responded “sometimes or often.”
If The Yes has a secret sauce driving this success, it seems to be in its user experience and approach to tech, which offers some key differences compared to others.
Stitch Fix, where Bornstein was once chief operating officer, uses questionnaires and games to drill into what a shopper might like, so it can show it to them — even if the person isn’t actually searching for it. Others fish around for the insights using historical data, like search histories, purchases and bookmarks, or some combination.
In contrast, The Yes’ namesake yes/no model was built to be fast, lightweight and fun. There’s just enough interaction to be engaging, and the curated assortment seems never-ending, always offering something new to see.
According to Aggarwal, who also serves as The Yes’ chief technology officer, the user experience is a top priority for the company. And it was able to focus on that, because it didn’t bog itself down with developing all the tech from scratch.
Instead, it relied on numerous Google Cloud tools, including Pub/Sub, Spanner, Kubernetes Engine and Cloud Vision API to build a system that could handle a large volume of data, both from customers and product data from the brands. Machine learning was key to the equation, so it could spot trends in real time and allow the company to make responsive decisions.
The startup essentially aimed to off-load whatever it could to the machines, so its people could focus unencumbered on areas that are uniquely suited to human beings — like the company’s “fashion taxonomy.”
According to Aggarwal, it’s a massive effort to organize, categorize and tag products. But it’s more than just slapping a few basic descriptors to garments.
“We create a lot of our underlying artificial intelligence models through intelligent labeling of the data. And that’s where the human expertise comes in, which is really critical,” he explained. “The models are trained by taking products and getting them labeled by humans, who tell us what the style of the product is.”
This is the sort of development that music streaming services like Pandora invest deeply in — how to identify genres of music, or categorize them by moods and other attributes that may not be easily understood by technology. The example in fashion, for instance, could be two flowy skirts with similar characteristics but appeal to different tastes and shopper preferences.
A selection of skirts, as seen in the app. Courtesy image
The work involves drilling down into a large range of details and nuances. So far, The Yes has incorporated 2,038 style dimensions in its fashion taxonomy.
“Then the algorithm not only [has] the idea of styles, but it can really scale this out over hundreds of thousands of products,” he added. “They also learn an effective ‘meta level,’ the nuance of style. So they learn what makes products similar or different.”
This mix of cloud tools and internal development has allowed The Yes to prioritize its efforts to move quickly and grow rapidly.
“I think it’s the best example of where all of retail wants and needs to go,” Carrie Tharp, Google Cloud’s vice president of retail and consumer, told WWD. ”How do you innovate faster? How do you take an idea to reality in a matter of months instead of years? How do you get tested more constantly, between your customer, your product? And what should your assortment look like? And what are you learning, in insights from that AI and ML about your customer, pricing, product, etc.? That’s what we need.”
That’s apparently what retail, in general, might need. According to a Harris Poll recently commissioned by the tech giant, less than half of retail executives globally were confident that their companies were properly equipped with the right tech tools for business continuity in the early stages of the pandemic, at 43 percent. And only about half of the executives believe their company overall is very prepared to deal with the shifting retail landscape stemming from the pandemic, at 51 percent.
This comes after a year in which 47 percent reporting that their businesses are accelerating cloud adoption — a group that includes Tapestry, which will migrate to Google Cloud through a multiyear agreement, Google said Thursday.
The holiday season has fueled some optimism, with 62 percent saying they believe their companies overall are very prepared to deal with the increases in consumer traffic. But 94 percent said COVID-19-related worries still keep them up at night, including supply chain issues and fulfillment.
Tharp believes The Yes is a shining example of how to effectively use cloud tools, especially in this environment.
Meanwhile, the company continues to look to new technologies, and not just at Google. For instance, it uses Apple’s App Clips — which allow small snippets of an app feature to run without having to download or launch the full app — to allow the sharing of guest lists with friends.
The Yes is also looking at the iPhone’s new LIDAR sensor, which allows for more accurate measurements using the phone’s camera. Such tech has major implications for things like digital fitting and size recommendations, which are areas of deep interest for Aggarwal and Bornstein during the pandemic and beyond.
Source: Read Full Article