All the Ways Zappos.com Leverages AI — And How the E-tailer Determines When to Get Humans Involved

For the last year, artificial intelligence has been the hottest topic not only in the tech space — but as an ongoing theme in the retail industry. While staying up to date on AI is not only a must, meaningfully integrating it into existing platforms has been a challenge for most.   

Notably, retailers can use the technology to enhance customer engagement, including shopping suggestions, personalization and customer service chats. More advanced integrations, like virtual try-ons, are also becoming more prevalent.

Zappos is no stranger to the world of AI and innovation to aid its customer engagement. In 2019, the Amazon-owned footwear and clothing company was a trailblazer with its self-taught AI algorithm that helped rebuild its search engine, increased customer personalization and picked up trending keywords and phrases.

Nigel Storey, chief technology officer at Zappos, sat down with Footwear News to open his playbook for integrating AI, enhancing personalization and using new tech tools to advance inclusivity in the fashion and retail industry and more.  

Footwear News: How has Zappos integrated AI into the company and within its e-commerce platform to improve customer engagement?  

Nigel Storey: Starting as a digital-first company in 1999, Zappos has always been at the forefront of technological innovations. Our company has always been “customer-obsessed,” and our mission is to provide excellent customer service across all areas of the business. Our team is constantly working hard to improve the customer journey and AI has been a key component of that innovation.   

There are varying degrees of AI technology available for companies to implement. At Zappos, we utilize a series of technologies that help emulate how a human being would perform a task. This includes generative AI to capture customer context and answer their questions quickly.  

We are also exploring the idea of a conversational AI chatbot to help customers looking for shopping suggestions, size recommendations or assistance with their past orders. Overall, AI can better serve our customers at scale. This ensures they have a more seamless and frictionless experience by saving them time, energy and effort.   

In what ways is Zappos leveraging AI to enhance personalization in the customer experience? What are some specific examples of these AI-driven initiatives?  

N.S.: At every stage of the customer journey, there is an opportunity to utilize AI to assist the customer. Think search suggestions and complementary product recommendations where our technology allows us to recommend various clothing options based on the shoes in a customer’s cart, analyzing customer sentiment, marketing communications and even virtual try-ons.   

There are multiple examples in the Zappos app experience of using Machine Learning (ML) and augmented reality to create a virtual-try-on environment. Customers who try features like virtual-try-on return about 30 percent less than customers who don’t.   

How can businesses utilize AI to enhance the data they receive and utilize it more effectively?  

N.S.: It all goes back to AI’s ability to process data. It also creates natural product associations, making it easier for us to harness and better assist customers online or through our product managers.   

Our team was tasked to create a series of automated dashboards. At first, this wasn’t something our principal engineers were closely familiar with, so they used AI to help them familiarize themselves with the technological landscape and write code to create those automated dashboards.   

What are some of the challenges businesses can face when integrating AI technologies into their existing systems? And how can they be properly addressed as we move into a digital-first age?  

N.S.: One of the biggest challenges is how AI shows up authentically to customers, how it identifies itself and how customers engage with it. We’ve all had that experience of using a chat feature to quickly get answers and then realized we were not speaking to a human and then things start to go downhill.  

We can avoid this by always having a path — similar to customers being able to dial 0 from a phone when dealing with an automated phone system — and using that to get and interact with a person.  

We have to include similar contingencies for chatbots, so when a customer types in something like “representative” or types the word “help” more than a certain threshold of times, they should be immediately routed to an available human representative because they’ve pushed through the failure point of what AI was capable of helping them within that interaction.  

How can companies use AI to make the fashion industry more inclusive?  

N.S.: As we work to ensure that everyone feels welcome in Zappos, we’re careful. There is an acknowledgment that AI still needs to be developed to ensure it doesn’t have any inherent coded biases that perpetuates non-inclusive fashion practices.   

We’ve used AI to ensure that things like adaptive products are recommended when needed based on previous purchasing behavior.   

How do you foresee the future of AI in Zappos and the broader e-commerce industry, especially as we are seeing customer behavior shifting towards more personalization?  

N.S.: We leveraged AI to develop safety and training videos for employees whose companies are part of our Zappos at Work (ZAW) program, which is a provider of safety shoes.   

We helped one of our customers roll out sneakers as a corporate gift to its employees, and we used Generative AI to enhance a training video (via subtitles, generated images and generated graphics) for their employees. We pushed the bounds of this and had AI generate the content in multiple languages, such as Spanish.