Reposted from Fashion & Mash
By Andy Narayanan, Vice President, Intelligent Commerce, Sentient
What’s the biggest difference between shopping online and shopping at a store? Online gives you the convenience of shopping in your robe, getting products shipped directly to your house, browsing an endless aisle of choice after choice, the ability to price check tons of retailers on the same dress, and a whole lot more. In a whole host of ways, shopping online is just, well, better. But there’s one thing that e-commerce sites have struggled with for years: the personal touch.
For many of us, when we head into our favourite stores, one of the best parts is interacting with a great salesperson. And what exactly makes a great salesperson? They get us. A great salesperson listens and understands what you want. They know which brands run a little small. They can intuit what your style is based on how you’re dressed and the clothes you’re trying on. They can head to the back to find you something they know you’ll love.
This, as we mentioned, is notably missing online. Instead of great salespeople, we have a search bar and little checkboxes to click to browse the aisles. We’re left browsing that so-called endless aisle in hopes of finding something we like. In essence, we’re trading the personal touch for convenience.
But new advances in artificial intelligence (AI) are changing all that. They bring far more personalisation than what you see at your typical fashion retailer and promise massive benefits for both retailers and their customers. Which, of course, is exactly how it should be.
So let’s get back to that online shopping experience for a second. You know the one I’m talking about. After all, most sites, when you get right down to it, have very similar interfaces. You’ve got a search bar, some facets to help narrow your search (those checkboxes on the left with options for brand, colour, size, price, etc.), and a grid of product images. If you don’t like what you see, you can click through the page numbers on the bottom or click into a product detail page to find more. And really, that’s most of what shopping online actually is.
But with AI, it’s different. Because AI can understand the product images themselves, it allows for a whole different kind of shopping.
A smart AI – like the one my company Sentient makes – looks at an image in hundreds of vectors. That means it can identify things that are tough to describe, like the placement of a logo, a certain kind of fringe, the height of a heel in relation to the rest of shoe, etc. But that’s not what really makes the AI feel personal to users. What does is how the AI reacts to their behaviour as they shop, in the moment.
So say a user starts with a normal search for a red dress. Each time she clicks on a dress to check the price or look at the product detail page, she’s sending the AI a signal. And the signal is simply that she’s interested in the product. What makes things personal is that the AI actively figures the similarities between the products the shopper is looking at. Is it a particular shade of red? The length of the dress? The scoop of the neckline? And as it’s learning what she wants, the AI can start suggesting dresses that fit her browsing patterns, not based on retailer metadata or purchases she’s made before, but from just the couple of clicks she’s made in the past few minutes.
That means, effectively, that an AI can figure out preference and style for that user. It knows what she’s looking for in a red dress, not just that she’s interested in a broad category of red dresses. In other words, it finds the red dress, not just an endless aisle of red dresses.
For retailers, implementing something like this is actually quite easy. It sits on the front-end of a site (no backend integration is necessary) and the AI can be specifically trained to their catalogue. All it needs, for example, is product images. After that, you can use the AI in your existing user flows, product detail pages, recommendation pages, you name it. It’s really up to retailers as to how they want to leverage it.
And the benefits are tangible. AI can help with important metrics like average order value, add to cart, and more, but one of the more interesting proof points is that AI helps expose the entirety of a retailer’s catalogue. That’s because, instead of using old recommendation systems (stuff like “users like you bought this”) or giving primacy to items that are already popular, an AI can look at the images themselves and recommend products that a user may never have found, because it was buried on page 40 or was from a brand they didn’t recognise or because the manufacturer didn’t give a retailer the right metadata.
In fact, our first customer surfaced a full 92% of their products in the first month they implemented. Which backs up a key thing we should underline here: AI really does know your entire catalogue. And as it learns, it helps a shopper find just what they want.
Other folks are taking different approaches to personalisation, of course. Chatbots are having their moment. Sites like The North Face have implemented a sort of Q&A flow that uses real AI to suggest products. A while back, Victoria’s Secret used a non-AI powered questionnaire to help users find the sizes and styles they liked. But what excites me more, is personalisation that adapts and reacts to buyer behaviour in the moment. One that learns style, intent, and preference as users browse. One that gives a shopper access to a retailer’s entire inventory.
AI learns, adapts, and gets the customer. It figures out what they like even when they might have trouble articulating it themselves. AI understands intent and style so that shoppers can stop scrolling through page after page of red dresses and instead, in just a few clicks, find the perfect red dress, just like a great salesperson would do. That means getting the key benefit of brick-and-mortar shopping without having to leave the couch. And that’s the sort of thing that wins you customers for life.