
In today’s fractured consumer landscape, one size does not fit all. To break through the marketing noise and the glut of options consumers are faced with, brands are under increasing pressure to offer personalized products and experiences. It’s a task that requires a more nimble supply chain, smarter merchandising and a customized sales approach.
At the American Apparel & Footwear Association’s Executive Summit Thursday, fashion tech companies Stitch Fix and Avametric discussed how they’re working to solve this challenge. Speaking during the “A Million Markets of One” panel, they provided insights into how their technologies are enabling brands to deliver desirable product, and they touched on the sometimes unexpected results that arise when using technology to aid in fashion design.
In the Stitch Fix model, shoppers don’t actually shop at all. Using a mix of stylists’ recommendations and computer algorithms, the company sends users boxes containing five items each month, which they can opt to purchase or return.
The retailer, which raised $120 million in its initial public offering, counts 2.4 million users among its clientele. Stitch Fix credits its popularity to the tomes of data it amasses on each user, which it analyzes to (ideally) deliver on a wardrobe full of goods that consumers are happy to receive sight unseen.
When asked how the Stitch Fix team can nail a subscriber’s tastes every time, data scientist Daragh Sibley was quick to point out that that’s not actually the goal. It was an interesting thing to say given the time and effort the data scientists and stylists pour into designing or buying for each curated box. Aren’t people signing up for this service to get looks they’ll love?
“Early on, we want to send you things you really love. Then we might want to help you push your boundaries,” he said, describing the value of that ah-ha moment when a customer tries on a piece they were initially skeptical about only to find that he or she loves it. “That’s the experience we can create that’s really sticky and builds a fantastic relationship.”
Sibley said those moments of discovery will further the company’s ultimate goal of creating a long-term relationship that simply delivering the expected couldn’t achieve.
In addition to buying wholesale, Stitch Fix also uses customer data and past product performance to design its own goods. This is where AI kicks in to crunch all the information about individuals’ likes and dislikes, their relevant demographics, like the weather where they live and the feedback people like them have provided over time. Then out spits design ideas—some of them inspired and others less so.
Case in point, Sibley recounted one computer-aided design that illustrated why the company’s human designers won’t be losing their jobs anytime soon. Apparently, based on data, a blue blouse with green polka dots seemed like a good idea. Though the product development team was skeptical, the top went into production. And in fact, the item sold pretty well. However, it did elicit some interesting feedback from some, most of which included the word “clown.”
The challenge is that the computer can model an almost infinite number of blouse designs, many more than have actually ever been sold, so there’s no sales history from which to predict how they might perform, Sibley said. “So, we can sometimes make recommendations about combinations of things that don’t make sense,” he said. Hence, the clown blouse. “This is the kind of thing that algorithms can’t understand because they hadn’t seen polka dots in these particular shapes and colors so it was extrapolating outside of the information we really had and created a questionable combination.”
Avametric is approaching the solution to the current fractured, demanding consumer from a different angle. The tech company teamed up with Gerber Technology to aid in the design process by providing cloth and fabric simulations and 3-D apparel visualization. Integrated into Gerber’s AccuMark 3D software, the goal is to streamline the time consuming, inefficient product development process.
“In most industries, designs start with a 3D model, whether your designing cars or cellphones. In apparel, you start with a drawing and you don’t see it in 3-D until you stitch it together and put it on a fit model,” said David Macy, vice president of product at Avametric.
The result, he said, is an initial sample that’s often off the mark, necessitating many more rounds of sampling. With 3-D models, the company envisions brands being able to cut that process in half—or at the very least, eliminating that first sample. Wit a shorter development process, apparel brands should be better able to nail trends, resulting in higher sales.
And when the products are available to the consumer, Avametric’s 3-D models can be employed again. At this stage, they’re used to represent the clothing on a model for the brand’s site. “Rather than a consumer seeing a photo of one model in one size, they can say here’s some information about my body size, show me an avatar that looks like me and let me try on different garments and see how they fit,” Macy said.
Avametric uses a battery of information, including height, weight, shape and the sizes consumers typically wear in popular brands to create their online likenesses.
When asked if body scanners would be more accurate, Macy admitted that scans are more precise but the challenge is, shoppers just aren’t into them.
First, he said shoppers see them as a hassle since a scan requires the customer to either get undressed or wear tight-fitting clothing to get measured. Second, people don’t love the idea of their measurements floating around in the cloud. But the real challenge was something the Avametric’s team couldn’t have foreseen before they ran a trial: apparently, there’s a bit of a disconnect between what people think they look like and the body they actually present to the world.
“It’s really hard to compete with people’s image of themselves,” Macy said. “So, when people do actually see an accurate 3D body scan of themselves, they quite often look at it and say ‘no, that’s not me.’”
So, while scans might be more accurate, they’re also more advanced than the consumer is ready for. And Macy’s team has found that the avatars the company creates with user-supplied data are effective enough. Thus far, Avametric said it’s seen online conversions increase by as much as 400 percent when shoppers use its tool. Though the company doesn’t have the data yet, it expects to be able to report a similar positive effect on return rates.