Why Wayfair isn’t ‘dogmatic’ about its AI usage

That was the company’s message at the annual Shoptalk Spring conference, during a mainstage session with the company’s co-founder and CEO Niraj Shah, and during a separate interview that its chief marketing officer Paul Toms gave with Modern Retail. 

Toms says Wayfair likes to be on “the leading edge” when it comes to AI. Not every use case is a hit, Toms acknowledged – he said Wayfair has gotten messages or comments from customers on social media decrying the use of “AI slop.” But he said that that’s OK. 

“I think the important thing is to make sure that you’re kind of keeping the pulse on the total audience,” Toms said. “You’re understanding the signal from the noise. You’re being respectful of those opinions and how you receive them and react to them.”

Wayfair of course, is now using AI for creative generation, and also for QA processes, to name just a few examples. But Wayfair has also sought to roll out new tools that use AI to aid in the shopping and discovery process.

Last year, for example, Wayfair rolled out a new AI-powered discover tab that uses AI generated creative to help customers visualize what different might items might look like paired together or incorporated into a certain theme. 

In an interview with Modern Retail, Toms spoke with Modern Retail about his philosophical approach to AI usage, what kind of feedback it has gotten from shoppers about its AI tools, and more. This interview has been edited for clarity and length. 

I wanted to ask you about AI because it’s a hot topic. Wayfair is an interesting company because you also have a lot of AI shopping tools yourself. Can you give me an overview of where Wayfair is at, what different AI tools might shoppers encounter in the process of shopping through Wayfair? 

“AI feels a lot like what the company felt like 20 years ago when we were just starting – technologies that allow you to do things that you could never imagine doing before

We launched a Discover tab a little less than a year ago, I think – the concept there being, I can now generate thousands or hundreds of thousands of inspirational images that allow you to find exactly the look that you’re looking for, and then immediately shop those products and have them delivered to you. And that might sound simple, but three years ago, to do that at that scale, you’d have to be running a dozen photo studios full-time.

In our category, and where the value is for the customer, it’s in that inspiration step. The customer knows what they want visually, but they don’t have a way to describe it, right? There’s a navigation step that hasn’t been easy for anyone shopping in these categories for, you know, decades that now we feel like we’re on the path to really unlocking that, and unlocking in a way that’s very specific to Wayfair.”

Are you actively talking about [the Discover Tab] in your marketing?

“Certainly we do have some amount of budget that we use for that, [but] it’s actually more just about the function of it. It’s about coming to Wayfair for an easy shopping experience, and it’s less about launching a new product.” 

What is your sense is of – how do you feel like the Wayfair customer feels about AI? And I recognize you probably have a lot of different cohorts. But do you feel like your customers – do they like using AI? Are they using it in their shopping process? 

“It’s very, very dependent in every case. At the same time that I’m sitting here telling you how effective it is and how folks really get a lot out of the Discover tab –  we’ve launched plenty of other things, including on the Discover tab, that customers have absolutely rejected, right?  I would also say that’s a moving target –  how folks receive this today, it will be different than how they receive it tomorrow, both because of attitudes about AI, but also how the technology evolves. It gets better. 

So to us it comes back to – how do I make sure I’m grounding myself in the customer’s shoes? How am I building things that make their lives easier? How am I doing that in a way that they’re willing to receive and use and not reject? And if you’re close with your customer, you should, you know, be able to sort of stay grounded there, and honestly, a lot of that becomes pretty self-evident along the way.” 

When you said, “We’ve tried things on the Discover tab that customers have rejected,” what comes to mind? 

“I love this example, but when we launched the Discover tab, the images were just generated from the ground up, they were kind of imagined tip to tail….And then our way to commercialize that for customers was to then show them things that look like the products and the images and say ‘inspired by the products and the images.’ But customers were like, ‘I do not like this. This is AI slop. You sent me something imaginary. I fell in love with it, and I can’t have this. So, like, what value does this add to me?’

So we went back to the drawing board, and the technology underneath also improved. We built the same thing, but now every item in the image is shoppable, like is accurately reproduced in the image to spec, to scale, alongside other products in the image that are also accurate. And customers loved it.”

I’m curious, is there anything you’ve told your team, “I don’t want you to use AI for this?”

I would frame it a little bit differently. In general, our mindset is going to be one of being bold and being on the leading edge. I think what’s important is that we’re humble, that we listen, that we’re willing to be wrong along the way. It’s about really staying close with the customers on what doesn’t work.

[With] customers – how they will respond, is definitely going to change, like you and I can probably recognize an AI [generated] face better today than we could six months ago. And I think that’s only going to continue. 

That uncanny valley is going to get smaller. And so I think for us, it’s more about not being kind of dogmatic about things, but being curious and being honest with ourselves about how people and how users and potential customers are receiving things.” 

Have you seen any comments from customers on social or through other channels related to AI usage? I talk to so many brands, and you get different anecdotes from each one, but I was talking to a brand yesterday… and they put out this ad campaign, and got a lot of comments on social media from people being like, “I hate this. This is AI.” It wasn’t AI, and so then they had to explain to people that it wasn’t AI. 

“Sure – there will always be that. Not [everything] that you do is going to be loved by everyone. I think the important thing is to make sure that you’re kind of keeping the pulse on the total audience. You’re understanding the signal from the noise. You’re being respectful of those opinions and how you receive them and react to them. 

One of the things that we’ve seen be successful – not just with our company, but I was talking to this CMO at American Eagle –  is some of these fantastical AI creatives that we put out. You know, you just have to set them in the right context. You have to declare that they’re AI and that you are being silly, right? Some of it is also how you frame it with your customers.” 

For the retail industry as a whole –  the way that they talk about AI, are there any ways in which you’d like to kind of see the conversation change?

“I actually find in retail folks are very hesitant. And I think my advice certainly would be, I think they’re probably being too conservative, you know, because these aren’t one- way doors. If you make a mistake, you can always stop, right? And you can revert changes. So to me, it’s actually – I’d like to see a little bit more ambition. I’d like to see a little bit less conservatism and more experimentation. I think that is how businesses grow.”