Nice outfit! Thanks, an algorithm styled it for me. Controversial? Maybe. If you ask a fashion editor (like I was for most of my career) it might all sound a bit suspect. Not to mention worrying. The same might be said for brand merchandisers and online content creators. They’d be forgiven for asking you and your outfit styling algorithms to politely step aside from their carefully crafted on-brand creations. How do you quantify creativity after all? And what about style? But that’s just it: when it comes to the power of AI and automated outfitting it’s not about quantifying creativity, it’s about scaling it. Outfitting is an important part of how a customer shops and as fashion consumption moves increasingly online, being able to scale a brand’s creative output by showing multiple, personalized outfits to each customer is a lucrative way to blend the online and offline experiences.
Storytelling: Outfitting Helps Brands Tell and Sell the Next Chapters in the Fashion Book
As we return to the world of real life living and shopping — away from the constraints of lockdown — overnight shop windows have been transformed from soul boosting aphorisms to mannequins decked out in all manner of seasonal style. Outfitting has always been the first port of call when it comes to telling and selling the brand story. From shop windows and advertising campaigns, to in-store merchandising, social media platforms and online stories detailing “what the model is wearing” on PDP pages, brands have been showcasing their products through outfitting right from the get-go. On the simplest level outfitting puts an item into context. It creates a story around it.
But this approach only allows a brand to tell the first sentence of the story. And it’s the same first sentence told to each and every customer. Automated personalized outfitting unfolds the next chapters. In the small but lucrative online space of a brand’s PDP page, for example, where you have the shortest amount of time to inspire a customer, showing multiple personalized outfits for one garment ticks so many boxes.
- There’s the inspiration box, showing how a customer can wear a piece styled differently for various occasions. Outfitting is the online browsing minus the strolling around the store factor.
- There’s the personalization box – showing a customer how to wear the jacket that is part of a trouser suit but styled only with skirts as she’s expressly stated: Never show me trousers.
- There’s the sustainability box showing a customer how to style a new garment with clothes already owned. That also ticks the “I have a cupboard full of clothes but nothing to wear” box.
Which Came First: Stylist Outfitting or AI Outfitting?
Stylist vs AI is a lot more delicate than the existential conundrum of the chicken and egg question. Quite simply, effective first class fashion AI cannot exist without the input of fashion expertise and creativity as a starting point. Fashion is not static. Just when you think AI’s cracked the t-shirt, straight leg jeans, trainers and oversized blazer look, designers, politics, social media and cultural influences will come along and throw you a fashion curveball. And it takes a human stylist to notice and understand all the issues and adapt the styling of outfits likewise.
Styling algorithms are not here to replace anything. They are here to scale a brand’s creative output. I’ve styled many fashion shoots in my time and it would be impossible to take a retailer’s inventory and style every variation around each SKU as well as every variation tailored to each customer’s preference. AI however can build on that original styling from IRL stylists by scaling it and most crucially keeping the style output on brand. For a retailer with a large inventory such as John Lewis, Dressipi can autogenerate 100 million personalized outfits each night, only showing outfits with product that is in stock. What this also means is that every garment in a brand’s inventory gets outfitted and loved.
In the early days of Dressipi, outfits were handpicked by stylists and shown to customers via weekly emails. It was bi-directional communication where customers could give feedback, helping to decode their expectations. As Dressipi grew, this approach became unsustainable and definitely not scalable. A better approach to generating outfits comes from a technical and data driven point of view, thus avoiding implicit human biases (as much as I love wearing heels with everything, I wouldn’t inflict this outfitting on everyone) and along the way overcoming several challenges in combining fashion and technology. We are not the only ones doing this.
Earlier this year Nordstrom announced the launch of its “Living Shopping” channel blending the virtual and in-store experiences with shoppable virtual events focused on outfitting from “How to Layer: A Virtual Styling Event” to Burberry’s styling event showcasing how to wear runway pieces with pieces from the rest of the collection. Fanya Chandler, senior vice president at Nordstrom explains: “Livestream Shopping enables us to stay closer to the customer with interactive and engaging experiences that allow for discovery, personalization and service at scale.”
In London, Farfetch’s store of the future initiative saw the launch of the new Browns store located in Mayfair. The physical and online experience and brand storytelling comes together thanks to technological innovation such as the interactive mirrors in dressing rooms that display items from a customer’s wish-list as well as showing how to style those items.
Don’t Ditch the Stylists
Don’t ditch the stylists. Only a combination of styling expertise and digital capabilities allows for a truly on-brand personalized approach to fashion outfitting. Which makes it not too incongruous that an algorithm styled your outfit.