Want to reduce returns? Avatars might be the answer

For more on avatars, sign up to receive the Vogue Business newsletter for the latest luxury news and insights, plus exclusive membership discounts. To become a Vogue Business Member and receive the Technology Edit newsletter, click here. While virtual try-on through augmented reality is sophisticated enough to realistically portray items such as eyewear, shoes and other accessories, it typically still doesn’t cut it when it comes to luxury fashion. But an alternative method has emerged. Fashion brands and retailers are experimenting with tools that let customers create digital twins of themselves that match their proportions and measurements, like a true-to-size avatar. These avatars are often based on customer-supplied measurements, images and body-type details. Instead of overlaying an item on top of an image, this approach fits digital clothing onto digitised versions of the customer’s body to help them visualise how they will look and ultimately select the best size. In February, Yoox Net-a-Porter (YNAP) began testing the technology on 90 styles from its house brand, Mr P, sold on men’s site Mr Porter. In November, Balmain began testing virtual fitting and styling, starting with seven ready-to-wear designs, five handbags and a boot. London-based Anthropics, a provider of photo-editing software, just announced the launch of Zyler virtual try-on for menswear, which is already used by Italian fashion brand Kangra and Italian bespoke menswear retailer Larusmiani. Others to trial similar technologies include Khaite, Canada Goose, John Lewis, Macy’s and gender-neutral clothing company Dapper Boi. The new tools all take advantage of recent advances in computing power, artificial intelligence and 3D software with results far more realistic than previously possible — a high bar in the world of luxury. In a release announcing its project, YNAP interim CEO Alison Loehnis noted that the company had been looking for user-friendly virtual try-on technology for a while. Balmain’s chief digital officer Simon Cottigny said in a statement that tech was finally at a point to “faithfully mirror Balmain’s unique signatures, patterns and designs”, and that the brand hoped to lower returns and reduce the brand’s carbon footprint while increasing sales; CEO Jean-Jacques Guével called it “a win-win”. An estimated 85 per cent of US apparel brands and retailers either use or plan to use virtual try-on tools, according to a 2023 report by Coresight Research, commissioned by virtual try-on tech company 3D Look. In June, Google piloted a tool that enables people to see items of clothing on a range of models and body types, while AI-powered virtual try-on and styling tech startup Zelig announced $15 million in Series A funding in November. Amazon has also just introduced AI-informed tools to help people find the right size. Better fit technology can help stem returns by making it easier for customers to find the right size before they buy. It’s a puzzle that retailers and brands are incentivised to solve. Return rates are one of the industry’s costliest problems: more than 24 per cent of online apparel orders in the US are returned, Coresight Research estimates — equivalent to around $38 billion in sales in 2023. The majority (77 per cent) of returns are made due to incorrect size or fit, according to a 2023 survey by supply chain management company Blue Yonder Group. This is compounded by the fact that many customers have got in the habit of bracketing, or buying multiples of one item to find their right size and returning the rest. The returns conundrum continues to be a major challenge for apparel brands, says Deanna Meador, CEO and co-founder of Couture Technologies, which provides fit tech to Dapper Boi. In research interviews with more than 300 brand representatives, many said virtual try-on is the “holy grail for online shopping experiences”, Meador says. “When customers buy and return ill-fitting garments, there is a financial, environmental and customer-confidence cost.” Previous attempts to create avatars often ran into issues with fidelity and accuracy, but many of those barriers have been lifted, thanks to recent wider tech advancements, says Christine Marzano, CEO of Bods, which provided the tech to Balmain and Khaite, and is planning to launch with a couple more big luxury brands this spring. Avatars, which borrow from the world of gaming, are unique in that they offer a 3D, realistic interpretation of the body, she adds. “No shopper has ever looked at just the front of a garment in a store or in a mirror at home and bought it.” Better fit, fewer returns The returns problem has forced retailers to take measures like reversing course on free returns. An estimated 81 per cent of merchants, including Macy’s and J Crew, are now charging a fee for returns by post, according to returns tech company Happy Returns. Even Amazon — which trained consumers to expect fast shipping and free returns — is scrambling to put the monster returns problem back in the box. Recent efforts include discouraging costly UPS returns; charging a fee for some returns; adding a “frequently returned” warning label for some products to prompt customers to do their research before purchase; and encouraging unboxed returns in physical Kohl’s stores for more efficient transportation. Early data suggests that digital twin fit predictors can increase customer confidence and decrease returns. Deepgears, which provided the tech for the YNAP pilot, says the brands that have used its tech see an average 25 per cent decrease in return rates and a 28 per cent in conversion on the items that offer the digital mannequin option. Companies that use Zyler see an increase in browsing time, higher engagement and a 10 per cent reduction in returns, according to the company. John Lewis’s rental service offers virtual try-on using Zyler: the retailer saw a 10 per cent reduction in online returns, and says that 38 per cent of sales happen after a Zyler visualisation. While the Balmain pilot is in early stages, Bods reports that there has been zero bracketing on the items that offer the 3D fitting tool. Among all brands who use Bods, which also enables customers to mix and match items, shoppers try on an average of five pieces per session and 25 per cent add their entire “look” to their cart. Early tests also show that retailers who offer the tech on some pieces see a 5 per cent reduction in returns overall across the site; this is expected to increase as brands increase the number of pieces that can be digitally fitted. Already, Balmain plans to grow its offering to include more of its “iconic pieces”, as well as new designs from the runway. YNAP plans to track the impact on returns over a six-month trial period, after finding that size and fit are the primary reasons for customer returns. Because the items offered using the technology must be first rendered in photorealistic 3D, it began by piloting more “evergreen” menswear styles. In addition to seeking better financial performance, brands are increasingly conscious of “doing well while doing good”, amid a big sustainability push, says Giulio Capodimonti, co-founder of Deepgears — and fit tech that reduces returns plays well into these goals. YNAP, for example, hopes that the ecological implications will appeal to customers along with the convenience of evading a return; it found that 83 per cent of its customers are trying to shop more intentionally to avoid sending pieces back. Couture Technologies’s research revealed that 95 per cent of those surveyed said that seeing an avatar and a fit prediction would make them more confident in making an online purchase. Couture Technologies uses a “numerical approach” to AI-powered fit predictions based on a customer’s measurements, gleaned via two submitted images. It also gamifies the process by showing circles of where the customer should stand when capturing the images. Brand partners saw no returns when customers had access to the tech, Meador says. Overcoming hurdles The space is still nascent, Capodimonti cautions, and while photorealism has come a long way, there is still room for improvement when it comes to garment fidelity — such as more accurate representations of how garments fit and stretch in certain points. Already, Deepgears can ingest technical specifications on garments and tweak textures so that the technology more accurately simulates the difference between denim and silk, for example. Technology is not the only limiting factor. Customers might be hesitant to share their personal information, or to put trust in a tech that hasn’t always been reliable. Many of the technologies require uploading intimate details historically reserved for the confines of a dressing room. To use the Mr P tool, customers provide their measurements, or provide their clothing sizes from other brands. Those creating a Bod on Balmain’s or Khaite’s websites, can input their measurements or submit two photos, in addition to manually tweaking details such as skin tone or visual hip width. To take advantage of Macy’s bra-fitting service (which used Fit:Match), store associates scanned in-store customers in person, before matching to a digital twin in an existing database. Those using Bold Metrics tech on Canada Goose, must enter their height, weight, age and bra size. Tech providers must also navigate if, and how, they store customer information. Those who create a digital twin using Bods can elect to sign in and use their avatar across retailers enabling the same tech. Marzano says that transparency and compliance with GDPR (General Data Protection Regulation) and CCPA (California Consumer Privacy Act) are key reasons why Bods doesn’t “white label” its technology (meaning disguise the tech provider). “The client knows exactly who they are working with even when they are shopping on different brand sites. If a user decides at a later time that they don’t want Bods or the partner brand to retain any of their information, we have included instructions within our ‘terms of service’ to allow users to ensure that all of their information is deleted.” The more burdensome returns become, however, the more customers may warm up to this type of personalised fit tech. Virtual shopping pushes like those on Apple’s Vision Pro could also help to normalise these types of AI avatars. For now, any change “needs to be incremental”, Capodimonti advises; brands are smart to start slowly to establish adoption and confidence in the tech. “We’re still in adolescence of the technology in terms of deployment. From an adoption curve, we are at the beginning.” But now that major retailers are starting to come onboard, he adds, “other people will start looking at it”. Correction: Updated to reflect how Mr P customers use the try-on tool. A selfie video is not required, as previously stated. 4 March, 2024 Comments, questions or feedback? Email us at feedback@voguebusiness.com. More on this topic: Virtual try-on is being hit by class actions. Should brands worry? Amazon rolls out an AI fit tool to reduce returns What we got right — and wrong — about fashion-tech in 2023

Make better business decisions

Make better business decisions