AI: the next fortune-teller
Cognitive systems are driving more personalized shopping experiences and helping unearth customer trends.
Until now, fashion has been largely speculative. Retailers have worked hard to forecast trends, pilot innovative products, and adapt to customer demand. However, they haven’t had the power to truly know what their customers want before they know they want it. That’s all changing. Now, retailers can harness AI as the crystal ball of fashion, using big data to get ahead of what’s next and bring designs to market faster than ever.
Predictive fashion is here, and it will allow retailers to demonstrate that they can move as quickly and intuitively as even the savviest customers and designers.
AI as a means of retail clairvoyance
Having a direct line of sight into trends that will arrive several months into the future is real power. Retailers are quickly adopting image recognition—both on social media and through cameras in trend-forward locations—as a way to understand the place their products have in consumers’ daily lives:
“The reality of the mobile and social media experience is that the pace at which change happens is quicker than ever. Consumer interest is changing very quickly, and retail at large needs to be aware and proactive about changing taste.” — Oliver Chen, Managing Director, Cowen & Company
Fashion startup Heuritech has built an AI platform that divines highly precise insights from millions of images on social media. Their first two partnerships are with Louis Vuitton and Christian Dior, both of whom use the technology to pick out their products on social media, understand how people are using them within their larger sense of fashion and personal identity, and hone in on potential hit trends.
With help from the Fashion Institute of Technology, IBM has built an AI search engine called Cognitive Prints that offers a balance of inspiration, authentication, and trend forecasting. Trained with 100,000 print swatches from winning Fashion Week submissions, the tool can use photos to search for similar items or even search based on highly specific styles, cuts, or individual elements. It can also design its own patterns, help identify trends, and ensure that designs are uniquely proprietary.
AI as the way to personalization and localization
Retailers can no longer hope to provide the same, uniform inventory across all markets. They need to have the data signals required to predict what will be fashionable in what geographies and plan local assortment accordingly. AI allows them to drill down not only to the local level but the highly personal level as well. And there’s immense potential here, given that customers are increasingly taking the reins of their own personal retail journey.
Having learned the costly lessons of uniform merchandising (which resulted in regularly unloading overstocked items at discounted rates), H&M now uses big data and AI to customize assortment by local clientele. Analysis of purchases in one particular store in Stockholm led them to overhaul their product mix to tailor it more towards women, emphasizing best-sellers like floral skirts, getting rid of most men’s products, and even introducing housewares and more premium apparel. After all, AI insights are only as valuable as a brand’s ability to effectively and strategically integrate them into the supply chain.
Other companies are moving full throttle into the customer-centric realm of predictive fashion. Epytom has created a Facebook Messenger bot that operates as a virtual fashion assistant. The bot takes stock of a customer’s existing wardrobe and proposes daily outfits based on a number of factors, including local weather and that customer’s plans for the day. It can even design complementary items to match their wardrobe.
GAN (generative adversarial network) is a type of AI that essentially generates realistic imagery. Researchers at UC San Diego assert that GANs could be used to project what customers want outside of available inventory. When customers send in their measurements for custom items—something certain retailers already embrace—GANs could rapidly (and inexpensively) respond with multiple different suggestions.
AI as the way to a leaner, keener inventory
Again, AI insights serve no function if they aren’t successfully delivered to market with careful consideration by retailers. But if brands can use AI to move faster and more predictively, they can be leaner as well, avoiding the pitfalls of stockpiling inventory. Tommy Hilfiger has found a way to bring collections from design to shelves in six months, rendering their “TommyNow” program three times faster than typical product rollouts. Joining with IBM and FIT, they have used AI to drastically speed up design, manufacture, and the supply chain.
The speed and agility made possible with AI-assisted design and production are deeply apparent. But one can also imagine the potential reduction in overall waste. AI can help us pursue a more exacting supply chain, with much less inventory bulk, but can it also help us lean more heavily on sustainable materials?
Move faster than fast fashion
Predictive fashion is not a fantasy, nor is it purely about telling the future. The tools and technology are here at our disposal, and they will provide retailers with a more responsible, adaptive, customer-centric approach to business.
As fashion moves faster and faster, how can you ensure that customers get the right fit when they visit your store? Read how AI is changing the future of the dressing room.