AI’s role in retail

Artificial intelligence has been enhancing the online shopping experience for some time: Put a new outfit in a shopping cart, and the store’s website suggests the perfect accessories as well as another outfit you didn’t know you needed. Start looking at a new flat-screen television and a reminder pops up that you also need HDMI cables and a TV mounting kit.

Now AI is increasingly making its way to in-store settings. The final installment in the NRF NXT webinar series explored AI’s role in retail. While robots and algorithms aren’t going to take over the world (at least not anytime soon), AI can help human experts do their jobs more efficiently.

Building customer relationships

At Neiman Marcus Group, AI, analytics and machine learning play key roles in helping the company build long-term relationships with its customers.

Styling and lookbooks were previously done manually — which is hard to scale across many customers and stores. When a customer makes a purchase now, AI categorizes order attributes such as size, color, pattern, brand, occasion, seasonality and lifestyle. These categories are entered into the AI environment, where they are overlaid with current customer data (such as purchase history).

Sales associates in stores feel empowered with better insights and a better understanding of their customers’ needs. For example, AI might provide sales associates with a repeat shopper’s age, lifestyle and style preferences.

“We could … offer the right content to the right customers at the right time,” said Hongpei Zhang, vice president, customer insights and advanced analytics at Neiman Marcus Group. “It’s really about that whole relationship, the journey and engagement.”

One size doesn’t fit all

The vision around The Yes was the idea of building a store around every user. According to CTO and co-founder Amit Aggarwal, most ecommerce experiences are “one-size-fits-all.” Using machine learning and AI, The Yes has built a store that is different for every user.

Noting the misconception that AI is meant to solve problems, Aggarwal said his company views it as a way to deeply understand customers’ changing needs, have a conversation with them and make the user feel understood.

While many believe that large amounts of data are required to use AI, The Yes believes that personalization is possible with small amounts of information.

“As we’ve built The Yes and as we’ve built our product, we’ve challenged those kinds of conventional ways of thinking about personalization and how AI applies to that,” Aggarwal said, “and we’ve tried to change that.”

Learning product attributes

With a large selection of products but limited brand signals that resonate with customers, machine learning and AI help Wayfair products find their way to customers’ homes. Wayfair uses artificial intelligence not only to learn product DNA (such as color or style) but also to better understand what customers are looking for.

Visual search helps Wayfair shoppers who have a hard time describing what they’re looking for, while augmented reality gives customers the ability to see items in their space — which helps them imagine owning it.

Fiona Tan, global head of customer and supplier technology at Wayfair, said the fashion and home categories are innovative facets of the retail industry — but they are challenging. 

“A lot of things don’t just neatly fit into a particular bucket,” she said. “A particular piece of furniture is probably a certain percentage minimalist, a certain percentage modern, et cetera. So, you’re really using AI models to train them on … the product DNA.”

Optimizing the supply chain

Amazon sees the potential for AI in back-office capabilities — which naturally translates into improving the customer experience. When thinking about the supply chain, AI can be used in forecasting, demand planning, assortment, allocation optimization and return optimization. It also can help determine the most optimal and sustainable packaging.

“When you’re shipping billions of packages every year and working with tens of millions of products, you can’t do it in that manual process,” said Steve Gurney, head of worldwide general merchandise at Amazon Web Services. “You really have to use the power of AI and the power of the cloud to do it at scale.”

Instead of packaging items in large Amazon boxes, over half of shipments are sent using flexible mailers. If the packaging provided by the supplier is sufficient, some items are sent with no additional packaging. Through artificial intelligence, Amazon hopes to use boxes and bulk packaging in only 15 percent of shipments by 2030.

AI is a tool, not an outcome

The sky’s the limit for AI; the technology has limitless capabilities in helping shape the future of online and in-store shopping. When asked about the future applications of AI, panelists suggested real-time personalization, data democratization, and utilizing vision and voice to power customer experiences.

However, Jason Goldberg, chief commerce strategy officer at Publicis Groupe and moderator of the NRF NXT webinar, stressed that AI itself is not the end.

"You’re setting your goal to use AI, and I like to remind people that AI is not an outcome. Artificial intelligence is a tool that we would use to achieve an outcome,” he said. “I think it’s much smarter to start an initiative to achieve a particular outcome. And certainly, it’s going to be the case that there are a lot of great outcomes in retail that can be helped by AI.”