Advances in personalisation, UX and AI are combining to automate creative tasks in retail
It seems hard to remember now, but there was a time when fashion retailers firmly believed that e-commerce would never cannibalise their in-store experience. Why, they asked, would the customer prefer shopping for products they couldn’t touch or try on, presented in boring little boxes, and without the benefit of guidance and help from store associates? It seemed a reasonable perspective, until customers made it clear that immediacy and convenience topped their list of priorities.
The upshot was a multi-year focus on building out e-commerce experiences. In the name of customer familiarity (and to emulate Amazon’s success), fashion retailers built out essentially the same features, using the same handful of platforms and tools. And when they finally looked up from their labours, they realised their online stores had little of the personality, creativity and distinctiveness of their physical stores. In successfully ‘Amazonifying’, they’d effectively stripped out much of what their customers valued off-line.
The ‘inspiration’ problem online
So, a new challenge has emerged in 2017 for retailers, especially in categories like fashion and home. With so many customers shopping online, how do you scale and digitise in-store experiences like outfitting, visual inspiration and style advice that engage the customer and distinguish their brand?
This is where the next generation of automation and AI are coming to the rescue. Take the question of inspiration, for example. Most customers shopping for clothing are considering two things: “Will it look good on me?” and “Is it worth the price?” In fact, 91% of customers surveyed said they’re more likely to buy something if they get styling advice on how to wear it.
Let’s consider our customer’s experience in store. There are mannequins, dressing rooms, helpful store associates, and even the shopping friend who can advise her just how to style that dress. That visual inspiration and validation make our customer much more likely to buy. And she might even pick up a few accessories on her way out.
91% of customers are more likely to buy something if they get styling advice on how to wear it
Online, retailers have traditionally had no way of providing styling advice for every product and every customer. Even if retailers had dozens of stylists working round the clock to create looks and assist customers – which most don’t – the sheer variety of product attributes and customer needs would make it impossible to support for long.
And so retailers have kept punting this problem away. After all, it can’t be supported manually and, of course, you can’t automate creative and subjective endeavours like styling…
Meet the AI stylist
Except, you can. In the past year, AI-driven outfitting algorithms created by my company Stylitics have been busy working behind the scenes at some of the largest retailers in the world. In a given month, they style and put together tens of thousands of products into tens of thousands of magazine-style outfit collages, which are displayed to millions of shoppers as they shop online.
Our technology uses a mix of trend and performance data, brand-specific styling guidelines, customer purchase and preference data, and a liberal dash of our secret ingredient to create stylist-quality looks for every product and customer – 200-500 times faster than a human being. Not only does our algorithm understand the different ways you need to pair each style – e.g. skinny jeans versus bootcut jeans versus boyfriend jeans; it can also factor in customer considerations like fit, and retailer considerations like inventory level, to create a visual outfit recommendation in milliseconds that it would have taken an expert stylist 20-minutes to assemble.
Our outfits have been shown to shoppers hundreds of millions of times in the past year, and not once has a customer said, “This was done by a machine”. In fact, in 100% of cases showing customers personalised outfits on every product page, and in interactions like post-purchase order confirmations, has increased sales by between 5-15%.
The future of creative work
I believe we’ll see more creative endeavours automated in the next few years, for better or worse. For example, most advertising (including video) will ultimately be AI-generated. And within the decade we’ll have at least one algorithm-generated top-ten hit song, and a top-ten bestselling illustrated children’s book. Companies like Stylitics are learning how to re-frame and re-think how humans generate and perceive creative work so it can be generated automatically and appear indistinguishable from human ingenuity. I don’t know yet what that means for our society, and for what we consider to be our unique value as human beings. But it’ll be interesting to find out.