Branding Without Limits: The Future of Retail Visuals with AI

Welcome and Introduction

Loreal Lynch: Hi, everyone. Thank you for joining. I'm excited to be here today talking about AI in the retail space for B2C retail marketing teams. And I want to just start with a little bit of an intro here. So Today's retail and B2C marketing teams are kind of in a tough spot. Interestingly, this is one of the segments that has grown the most for Jasper over the last year. And we think that the reason is that from what we're hearing, not only are they being challenged to deliver a higher volume of media and branded content than ever before, but there's a lot of pressure these days for these assets to be personalized and for them to have demonstrated roi. And this is an area where I can really help. I can help develop these images, scale the creation of these images and branded visual content. But the issue that we're seeing is that a lot of the AI technology that's out there today, particularly when it comes to generative AI, has really been focused on text generation. And ironically enough, image and visual has taken a little bit of a backseat. But we're seeing that that's really starting to change because arguably, visuals are probably the most critical component of retail and B2C marketing. So today we're going to delve into this topic a little bit more. We're going to talk about where visual media creation and branded content is headed in the age of AI. And I'm super excited to be joined here by Brian Godwin, who's an expert and a thought leader in this space. Brian, thanks for being here.

Brian Godwin: Thanks for having me. Very excited to be here. Thank you.

Guest Background: Brian Godwin

Loreal Lynch: Yeah, I'm just going to brag a little bit about your background because I think it's very cool what you've seen and you can add anything that I've missed, but Brian is currently the director of AI and visual media production at Wayfarer. But he's actually spent the last two decades leading creative technology teams and companies. And in that time, he's seen a lot. He's navigated the dynamic landscape of digital business. He's transitioned from the worlds of film and television now into the world of E commerce. And he is really at the forefront of tackling AI, machine learning, 3D visualization, all of this emerging technology head on, and to put that into practice in the work that he does every day. So anything else that you would add to that, Brian, based on kind of your journey?

Brian Godwin: Yeah, that's a great intro. Thanks, l'. Oreal. Appreciate it. Yeah, as l' Oreal mentioned, I am the head of Visual media production for the marketing team here at Wayfair that's inclusive of all manner of things, from photography to video, to 3D visualization to broadcast, if there is visual media attached to it for marketing uses. My team of amazing folks probably made it. But in addition, I've got a great team of researchers and artists and developers that help us really propel AI towards the future. Here at Wayfair, AI is really being sewn into the fabric of everything we do, and pretty much every initiative that everybody's trying to tackle has some sort of AI amendment onto it. So it's a very exciting time to see how, you know, all of these new tools for creating content, whether it's for entertainment, whether it's for retail, all just being supercharged and accelerated by all of the recent technological advances that have happened.

Loreal Lynch: That's awesome. Yeah. And I definitely want to dive into some of the specifics.

AI Trends in Retail Visuals

Loreal Lynch: But to start, we have this stat from recent research from McKinsey which says that 20% of retail leaders are actually making generative AI their number one priority this year, and they're ready to make significant investments here. So I just want to ask you, based on the work that you and your team are doing, what AI trends are you the most excited about?

Brian Godwin: Yeah, I'm surprised to hear it's as low as 20. I would imagine it would be much higher than that. And that people are kind of coming around down to this, being the next generation to propel content forward, you know, excited about all kinds of things. The great thing about being active in this space is that it changes almost daily and that there's a new technological advancement that takes you by surprise. I'm old enough to been through like a lot of technical revolutions, the dawn of the Internet and how, you know, visual effects in my previous career went from, you know, people building models by hand and painting matte paintings to all being done in computer generated imagery, you know, and I have never seen anything that moves as fast as what's happening today. So there's just, there's so much to be excited about. But, you know, the trends that I think everyone's kind of got their eye on, primarily hyper personalization is probably the number one, number two, branded content at scale. Because branding is a very subtle thing. It's a tricky thing. It's a very human thing. And when human capital is involved, it's hard to accelerate it at scale. But with regards to the personalization stuff, I think that the ability to create content dynamically and create content dynamically that appeals to Consumers. And whether that's in the space of trying to sell something or trying to tell a story, I think that personalization is really the thing that everybody across all fields are excited about. To be able to have content that really resonates with an audience. Again, it's not just about selling stuff. It's a very exciting time. We could potentially have content that appeals to what we dream about, what we want to see, the stories we want to hear, the things we want to buy, the stuff we want to wear. So that's probably my number one area of excitement there.

Personalization and Customer Experience

Loreal Lynch: Yeah, so like just diving into personalization a little bit because personalization has been something that we've all been talking about for, you know, years. But I think to your point, it's a little bit different now with AI. And so what do, how do you see sort of like personalization really changing and transforming the retail industry within the next like three to five years?

Brian Godwin: Yeah, absolutely. I mean it, you know, would be no surprise to everyone that there's a lot of data out there. Right. And there's a lot of data about you, a lot of data about me, a lot of data about everybody. And you know, this, this data, we kind of give it away in exchange for goods and services, but hopefully that data allows content to be served to you that is appealing to what you're looking for. Again, whether that's an entertainment aspect or a retail aspect. But within regards to retail, you know, I foresee in the next couple of years that ads are going to have content that's hyper targeted. Not just broadly content generated for like geolocation or general preferences, but you know, seeing a product that you're interested in in the context of a room home, a style, you know, anything that is your particular sort of vibe or look. The hope is that we can start to serve up that kind of stuff at the top of the funnel, you know, in a broad context to really kind of bring the consumer in. But then once we kind of land them and get the consumer into the purchase funnel to be able to help them visualize, you know, and see the product that they want to buy either in a virtual try on context, if it's garments and apparel, or for home, for us being able to see that product when in a context that's either similar or specifically their home, a photograph of their home to see what it's going to look like, you know, and particularly with, with E commerce. Right. Like if folks haven't been focusing on retail and physical stores though I do think there's quite A big swing going back that direction. But when the vast majority of the products that you sell on a platform, you know, you can't go touch, see and feel, having that context and personalization is really critical to understanding how that product's going to resonate within your own style, within your own, you know, within your own look and feel and preferences.

Loreal Lynch: Yeah, absolutely. So I think it's really interesting to think about how this can kind of improve the customer experience as well as the work that we're doing. But I want to actually back up a little bit into some of the challenges that this is solving. And I mentioned some at the top of the call on just the pressure to deliver more faster. It needs to be personalized, it needs to deliver roi. But if you can talk specifically at Wayfair around some of the challenges that you're having and where you're hoping that some of this new AI technology can help, would love to hear that.

Challenges of Branded Content at Scale

Brian Godwin: Yeah, sure. I think the biggest area that we have an interest in and that other companies too are interested, is branded content at scale. At Wayfair, branding is critical. I mean, for any brand, your brand is paramount. Being able to communicate a brand voice consistently is critical. As a platform, we have multiple brands, we have specialty retail brands, we have luxury brands, we have brands within the Wayfair ecosystem. And they all have an interesting and specific and very detailed voice that goes with it. And that voice is, you know, unique to not just copy, which is very important. We've certainly been partnering with Jasper quite a bit and quite effectively in terms of branded copy, but in terms of branded visuals. You know, we go through great lengths to merchandise, curate and pull together various suppliers, various products into collections in a way that feels like a brand, so that the consumer can land there and say, oh, you know, that's my style. I'm an all modern person. I'm a joss and main person. And that content has to have that consistent thread of that brand voice all the way through the content. And currently, you know, that takes a lot of effort and a lot of labor from very talented creatives, very talented marketers, very talented photographers, 3D visual artists. And, you know, we can do this much of it and it comes out great. And it's not necessarily inexpensive, but it doesn't allow us to operate on the scale that we'd like to in terms of producing that content with that particular brand voice in the tens of thousands or the numbers that we need to get to for advertising purposes for that hyper personalization aspect of it. You know, how do you keep the brand voice intact while still resonating with the brand consumer? That's a tricky pain point and hard to tackle. You know, like with, again with copy, there's a lot of tools out there, particularly Jasper. There are tools that allow you to work on your branded content, work on your brand voice, you know, upload branded guidelines and copy that is already approved and synthesize that into outputs that are dynamically generated. Doing that with visuals far harder, takes a lot more compute power and the ability to train generative models on a particular brand voice is tricky. It's certainly doable and the things that we're working on and I know others are working on, but that's one of the biggest pain points.

Loreal Lynch: That's challenging. That's challenging because if we think about generative AI, it's great at coming up with general outcomes. It can create images that are kind of along the lines of what you're talking about or what you have in your head, but tell it to create that specific thing. Now that's where it gets really challenging and it gets even more challenging when it's something just as subtle and coded as branding. So that's a tricky bit that I think everybody's up against right now in the retail sector. That is one of the things that I hear a lot is that kind of the early days of visual AI, you ask AI to create a blue frog on the Empire State Building, fine, you can do that. But when it comes to a brand like Wayfair or like you mentioned, one of your sub brands and needing to create very high quality visuals that are consistent and on brand, it's a much taller order. I know we have a team, Jasper Research Labs that's focused on trying to improve this produce research and open source research on improving these models. They recently released some research on shadows and the placement of shadows in visuals or on creating a reference image for brand and being able to build images off of that. So it is an area to your point that I think there's a lot of and innovation that that's happening here. But, but one of the, the questions that I have that kind of goes along with this is as the, the technology does improve and the adoption of AI generated visuals starts to increase, one of the questions and the concerns that I get from brand and creative teams is how do you maintain the creativity? Actually it's a question that just came in on the webinar. How do you not make things look cookie cutter, become a sea of sameness? How do you balance creativity with AI?

Balancing Creativity and AI

Brian Godwin: Yeah, I think that's a really great question and just super important to emphasize how important creatives are in this process. I think there's a lot of fear and concern about AI being something that will take over the creative process, but I really like to look at it as just another tool in the toolkit. You know, we've gone through various digital revolutions where we've gone from, you know, people in creative positions working with paint and markers and paper, drawing storyboards by hand and moving on to Photoshop. And Photoshop was going to put everybody out of work and, you know, film moving to digital, and that was going to put everybody out of work. And it hasn't happened. All it's done is actually democratize creativity further. Anyone with a good idea and get that idea across in a larger scale. And I think that that's, that's really critical for teams and for businesses to acknowledge and to really implement in their AI strategy. You know, it's really easy for folks who are looking at the spreadsheets and they're looking at, you know, the KPIs that we're held to and what we need to do and just say, hey, this is a great tool to get there faster and bigger and at scale. And it is. Right? But to the commenter's point, if you don't have great creatives at the helm, you know, making good templates, creating the brand standards, doing all of the real creative research to understand what the consumer wants and how to tell that story to them in a way that they understand, you are going to get a sea of mediocrity. And I think that just reinforces the point that, you know, we need thinkers, we need creatives, we need visual artists to be at the forefront of the process. And then Genai can be that supercharger of the engine that can take those ideas and really rapidly scale them up and get them out to consumers of the content. And I think that's really how you get to it. Even if we think about something like shadows that your team is working on, that's a very small detail that's very important. When we think about content that we're serving up on our product display page, for instance, we have, you know, what we call a silhouette of a product on white. Now, that silhouette is branded from the type of shadow, the quality of shadow, how dark the background is, how the product sits into that. And, you know, to be able to control things at that level of granularity, you need great creative teams interfacing with the science and research teams to Drive how they train the models to get that branded content just right. If you didn't have that upfront, well then you just get a generic shadow and it's not your brand's shadow. And even down to that detail, it does matter.

Loreal Lynch: Yeah, that's really interesting and I think what you're saying is right. And we, we see the same with respect to, you know, the AI is really meant to augment human creativity, but it does start with the human and with that, the human that has the idea that, you know, often it's these creative teams, but we often see that while the leaders tend to understand that and are advocating bringing on AI. How do you message that to your teams who oftentimes we see that teams are in fear. And again, this is another question that actually just came in. How do you communicate to the teams and to the orgs about the fact that AI adoption will not replace in house creative teams?

Overcoming Fear and Driving Team Adoption

Brian Godwin: Yeah, I mean it is a constant challenge and something that I think as creative thought leaders, we're always thinking about. I think the first and foremost thing is to get folks into the tools, you know, and get them playing around and having fun. AI is pretty fun. Like, I don't know that there are many people who have like, you know, fired up a mid journey or a dolly or even chat GPT and not had a good time. Right. Like it's, it's pretty fun to, to play with and I think, you know, bringing that spirit of exploration, curiosity and just saying like, hey, just, just get in there and play around, come up with ideas, see what it can do for you. I think people realize very quickly the advantages that AI have once they get past their fear and actually put their hands on the wheel. But they also realize the limitations and they realize, oh wait, this probably isn't going to take my job because unless I give it really good ideas, it's not going to put anything out there that is worth putting into the world. Some of my colleagues in the content team, they've been really great about how to drive this forward with teams and putting together things like AI Day. One of my colleagues came up with this idea and it's been great. We'll get the whole content team together and we put people from different teams onto sub teams and just say, hey, we need everybody to come up with a campaign or come up with an idea. And some people on the team, they're everybody from visual artists to marketers to tech people who maybe have never created a campaign. So here's a suite of tools go bananas and come up with ideas. And I think we actually had some folks from your team come and speak at one of these aids. But it's just such a great way to break down those barriers and make it fun and allow people to sort of demystify the fear of these tools and just realize, you know, yeah, it's a lot different than getting into figma or Photoshop or, you know, any of the tools that we use in our day to day. But it isn't that different. It's a means to the end, at the end of the day. And if you can get people to have fun, it's a little easier for them to adopt the tools.

Loreal Lynch: Yeah. You know, what's interesting is our data is kind of consistent with what you're saying, which is that there is a lot of fear before people are using the tools. And once they get in there and start to actually adopt them, we have seen that employee survey scores actually increase because employees tend to be more engaged, they're learning a new skill, they're excited about work because there's a lot of burnout as well. And so we're seeing that AI and learning a new skill is kind of an antidote to that. So I love that that's what you're seeing within your own team as well.

AI Workflows and Staying on Brand

Loreal Lynch: Another question that that has come in here is just kind of going back to what you were saying on it being challenging. You talked about this as a challenge of, you know, staying on brand, staying consistent with images. A question someone had the question of, like, how do you do that? But I had kind of a more direct question which is, can you actually describe a workflow that you have internally where you're using AI and just walk us through what that looks like, how you stay on brand, how you implemented it, anything that you can talk about?

Brian Godwin: Yeah, I can talk about some things without getting too heavy into secret sauce specifics. But in terms of visuals, we're augmenting a lot of existing content with AI to amplify its use and reuse. We also use a lot of AI to improve and edit our existing assets as well, as well as new assets. And this goes into the world of 3D visualization in a lot of ways, but also in terms of like photography and video. You know, what we are trying to do is be able to create an idea or basically like a canvas that says, like, here's the composition, here's what we want it to look like. This is what we're trying to make it look like in the end. And then using AI to Augment that into something that's either more photo realistic or even, you know, take something as simple as a photograph and turn it into something with video on the other end. And I think that's, that's an interesting trend that's really starting to pop up now is, you know, all of the video generation stuff that's out there is all pretty exciting. I would also say, you know, and I don't think this is a secret sauce thing, but a lot of folks are working on basically being able to take products and put them into contextual backgrounds and contextual environments. And you know, that plays into the hyper personalization space where, you know, you can take that, that existing asset and put it into a new universe that, you know is going to resonate more with the brand or resonate with the client to a certain extent. Like the, the personalization aspect is very similar to the branded aspect. At the end of the day, we are trying to train models to understand something specific in terms of a visual outcome. And that specificness can be a brand voice or a brand style book, but it can also be like your style or your home. And it's a similar type of approach in terms of the training of the models to be able to get to the specific outcomes.

Loreal Lynch: Right? Yeah, that's one of the use cases that we see with our retail customers, as well as the personalized background, incorporating a personalized background against like a product. So that's very cool. And then kind of along with that, one of the other questions that I had is maybe you can talk a little bit about. Gartner says that one of the biggest barriers to AI adoption is being able to connect AI to business value. One of the mistakes that we see is when customers start to use AI and they don't have a clear way that they're going to measure the impact. Can you talk a little bit about how your gold. And as you think about rolling out these AI tools, what are you looking to in order to measure success? What are some of the KPIs that you look at?

Measuring Success and Business Value

Brian Godwin: Yeah, you know, great question. I think that a lot of the KPIs are similar, right? It's just the amplification of the content that changes. You know, they generally tend to roll around, you know, speed, cost and quality. Like that's, that's always the big three when it comes to particularly any, any sort of, you know, media asset, whether it's copy, whether it's visual or otherwise. But you know, in the past it was kind of like, well, you got these three Main things, you only get one of them, you know, good, fast or cheap. You can't have all three. And I think that where we're trying to push, particularly with AI, is to be able to take, you know, good with a lot of human interaction. It requires really good upfront creative. It requires great artists, copywriters, all marketing. You can kind of nail that down and, but then like the speed and cost aspect, which are usually at odds with each other in sort of more manual workflows, those, that's where AI can really get in and actually get that brass ring of getting all three of those things, you know, fast, cheap and good at the end of the day because you can kind of shoulder that speed and volume part of it onto the AI. It's always going to be faster than any human in the loop function. That said, you know, even through that process, there's humans in the beginning, but still, you know, no matter how much computer vision we throw at our QCQA process, there's still got to be people on the end just making sure it gets out there.

Loreal Lynch: And have you had any issues with having to build a business case or prove it, or do you feel like the culture is such that it's encouraged to adopt these new tools and technologies?

Brian Godwin: Yeah, absolutely. I mean, you know, Wayfair brings a lot of rigor to all of our workflows and I think that that's a really healthy thing. We want to go fast, we all want to go fast. But when it comes to content and to branding and marketing, we want to make sure that we are putting creative first and we're putting branding first and that we're not just rushing visuals out there into the world that are either watered down brand. A lot of work goes into creating these content stories. And if we move a little too quickly, it's very easy for someone to just generate 10,000 images of something that isn't good. So still, creating business cases is really at the core of a lot of our discovery. Most of our generative AI starts out like any other proof of concept, right? Like we get in there, we sort of build a pilot, we propose assumptions, you know, we figure out what is our baseline, where do we want to get to, how are we going to measure that? I don't think that's anything new, it's just that the tools underneath in terms of where we want to get change out there. But I think that's still really critical to go through that crawl, walk, run process of, you know, starting out with a pilot, proving it out, building the business Case, making sure the budget and rationale are there, making sure that, you know, there are outcomes that are actually going to drive business impact. I think that's really critical to keep in mind rather than just like going bananas right out of the gate.

Skills, Upskilling, and Hiring for AI

Loreal Lynch: That's awesome. Now I want to jump back into talking about the teams. We talked a little bit about the fear. Once you get over that hump and you have more of a willingness to learn. The second issue that we see that's a barrier to kind of full adoption and scale is that there's often just a skills and knowledge gap. And if you think about the new workforce, particularly when it comes to creative teams and the use of AI, there are, I think, new skills that these teams are going to need to learn in order to harness these tools. Can you speak a little bit to how you're looking at skill sets now, particularly with, like, hiring new people or upskilling the existing workforce and what, what skills you think they need to have in order to successfully leverage AI?

Brian Godwin: I love this question. This is at the top of my mind all the time and something I work with pretty much daily with my teams. You know, I think there is that initial adoption we were kind of talking about. Make it fun, demystify it, get them in the tools. You know, it's kind of like, yeah, you can get someone in a car, you can drive them around. It's really fun. You see the landscapes going by or in the good day on the Sunday drive, but then it's like, okay, now you have to drive the car. Exactly. And like, that's. That's where. Where it gets a little more challenging. Right. And, you know, in. In the sort of spirit of, you know, basic AI is really easy to get. Generic outcomes that look good but aren't necessarily outcomes you're hoping to drive. There's a cliff that happens where like all of a sudden it takes a lot of talent and skill to be able to pilot these tools to get the outcomes you want. And, you know, what we're finding is folks who just have a nature of curiosity and to adapt faster. I don't think this is new to AI. I think it's just a creative technology thing that has always been the folks who have been successful, that they have a thirst for knowledge, they have an interest in tech in general, and they have creativity, they have ideas, and they have ideas they want to get done. And those folks seem to be the ones that persevere and adopt the technology in an actual useful way to be able to drive real outcomes. Because Once you're past something simple like just text to image, text to video, I mean, that's great. Like, describe what you want. You mentioned a blue frog in the Empire State Building. Great. But when you want to get in there and you're like, okay, we're building a brand campaign, we want to create visuals that tell a particular content story that's going to drive sales to this particular sector of what we're trying to sell. Then it gets hard and the tools get harder too. We're still in very early days when we're talking about AI tools to get really specific outcomes. Like, they're complicated, you know, it's not just as easy as a text box. You need to really get in there. And oftentimes you need to link together multiple modalities, multiple applications, multiple outputs, and how you chain those things together to get an outcome can be complex. But again, when we look at folks that we want in this sector for the future, I look for folks who just really enjoy thinking, systems thinking, who can really synthesize, like, new ideas and, and embrace them out of the gate where we don't have to do that. First part of proving to them that it's fun. They're like, I know this is fun. Just teach me how to get in there. And if you can't teach me, just give me a stack of tutorials. I'm going to rip through them and I'm going to get to the end goal really fast. And we're fortunate. We have a lot of folks like that on our team that have really dove in headfirst. And as fast as this stuff changes, you have to be curious, you have to be thirsty for knowledge, you have to want to get in there. Because if you're just looking at it as like, all right, I need to master this tool and I master the tool. Hey, sorry, that tool's out of date. There's a new tool and like, that's a, you know, 72 hour cycle or something. Ridiculous.

Loreal Lynch: Yeah, yeah, yeah, it's, it is interesting. And I, you know, I've actually been getting this question a lot as well. One of the examples that I've been using is that historically I've, I would ask in an interview cycle with a candidate for them to submit a writing sample to gauge how they message, how they, you know, their storytelling skills, their, their writing skills. Now, now I don't ask for that anymore because it can easily be AI generated.

Brian Godwin: Yeah, absolutely. Yeah.

Loreal Lynch: And so there are more questions to your point that gauge intellectual curiosity, that gauge, you know, their Desire to roll up their sleeves and learn something new. And tell me about an example about, you know, when you have had to do that before, when you've brought in a new technology or championed a new technology, those typ or learn something new on the job that you've taught to others. I think those are important skills to be looking for as we hire. And it's not necessarily experience with the tools because a lot of them are so new, but it's willingness to get in there and try them out and champion them because that's, you know, that's going to make a huge difference on our teams.

Cross-Functional Collaboration

Loreal Lynch: Another question that I had for you is beyond the team, the team adoption and kind of the team upskilling, what are the relationships like that you have with cross functional partners that are necessary in order to get some of these tools in place? Whether it be, you know, it or, you know, curious what other teams you have to work with on this?

Brian Godwin: Yeah, absolutely. And we're, we're a very cross functional company. And that's one of my favorite things about working at Wayfair is that everybody partners together and there are very few barriers. If you want to go have a conversation and learn about something here, you can go knock on a virtual door, anybody's office and get in there and talk to them. But that said, you know, it's, it is, it is, it's, it's really refreshing and it's a very cool environment in that regard, you know, but we do have various columns that operate together. You know, a good example would be the way that marketing works with content, works with advertising. You know, we all have kind of the same goal at the end of the day or some more stuff. But we need to partner together to be able to pull that thread all the way through the customer experience. We talked a little bit about putting together case studies and putting together business proposals. I think that that's pretty critical because not everyone's a visual thinker. Not everyone's going to be really clued in on something as subtle as the shadows of a branded silhouette. But they need to be able to understand what business outcomes that drives and being able to illustrate that in terms of results, how we measure results is really critical. So we start that way in general in terms of how we bring people in. But at the end of the, the results are really what matters. As long as you can illustrate where you're headed. You know, with a beginning hypothesis and then a conclusion of how a pilot went. You know, most of the time our stakeholders tend to Be pretty amenable to the ideas. What's really cool to see is how two years ago versus today, the appetite for these tools and workflows has changed tremendously. When I've actually been at Wayfair, I think it's three, might even be three years today or something crazy like that. But I was coming from entertainment and visual media for movies and television, and we had been working with Generative AI for years for special effects needs and for things like changing the faces on a stuntman, creating crowd simulations, all that stuff. It's all machine learning at the end of the day. And we've been implementing it for a while. And so when I got here, I said, hey, are you guys hip to, like, this Gen AI imagery stuff? It's wild. And you know, to be fair, like, I could see these really rough, ugly, stable diffusion one images and see the potential getting my stakeholders on board with that. And there's. I don't even know. I don't know what that is. Yeah, I don't see. It was really hard a couple of years ago, but now when I show them something like, let's, let's take Generative Video, for instance, and say, hey, this isn't perfect, but with a little investment, we get it there. Now they're like, really? That sounds great. What can you do? How much does it cost? How big can we go? How big can we get? And that enthusiasm and excitement has really just unlocked so much possibility in terms of how we can creatively at all of this stuff today, but also what's coming down the pike tomorrow, because we don't really know what that is yet.

Loreal Lynch: That's so interesting and so important, this culture of innovation and experimentation, just to dive a little deeper. And a question just kind of came in on this. So that makes sense for the overall onboarding of the tool and building consensus. But then when it comes to the actual technical rollout, do you have within Wayfair, is this something that you do within your own team? Do you have to partner with technical resources to get this onboarded or how do you manage that?

Technical Rollout and Training

Brian Godwin: Yeah, you know, it depends on the scale and scope of what you're trying to do, and it depends on the product. You know, we work really closely with the creative operations team that's really involved in terms of rollout of creative tech. I'm very fortunate to have an extremely talented tech team within my org alone. And, you know, it's. It's like five people that just when we talk about, you know, thirsty, hungry creative folks, those guys create all kinds of tools for all of us to execute with and they create them incredibly fast. So having that in my team is really a luxury. But then we of course have all of the broader technological orgs and when we get into something larger that might impact what we're doing on the storefront or on the website, getting those teams involved is really important as well. Once we do kind of pick something though, in terms of implementation and rollout, we do a lot of training. We do a lot of like one on one. We try to host, you know, like internal kind of like webinars or training sessions that we record and then disseminate. But we also try to target the user base too. Right. Like, I wouldn't go and try and roll out like a very complicated image creation workflow, you know, to someone who's coming from a photography background that's not their background, they're not going to understand it. So sometimes we'll take those workflows and skin them in a more simple way so that they can understand them and get value out of it, out of the gate. We'll train them on that subset, what the tech does. And then now We've got our 3D visualization team a little bit more tech savvy. We'll sort of unlock a little bit more for them, show them that stuff. And then we've got our engineers and real technicians, really strip all of that window dressing off and be like, hey, look, these are really just a bunch of little python modules that link together. You can create anything you want because you can understand how to put all this stuff under the hood together. So it really just depends on which audience in terms of how we roll it out. Training, we roll out with it.

Loreal Lynch: Cool. Well, that makes sense. I have a couple interesting questions that just came in and I think we have a couple minutes left. So let me go through these and then we'll, we'll wrap. But one, one of them is how much does Wayfair AB test? And do you think AI will eventually make AB testing obsolete? Generating and optimizing the perfect creative in real time for each audience.

Q&A: AB Testing and Brand Compliance

Brian Godwin: Yeah. So tons of a B testing. We're, we're, we're big on data. Like, it's a very data driven company. Was one of the things I was really excited about when I came on board here is like how much data we have and how much we respect leveraging that data for outcomes. It's very, very cool stuff. So. So yeah, that's a cool question. That's the hope in some way that like, you know, we can use different algorithms to like promote the winners and promote the losers. Everybody kind of does that in terms of like performance marketing, that kind of thing. But, you know, I just, I just don't think it would ever become obsolete in terms of like actual human testing. I think that humans are unpredictable. They change very frequently and their needs, wants, desires change and they adapt to the tech. You know, we're an adaptable species, we're always evolving. And I think that like, you need humans in the end of that loop just like you need them at the beginning of the loop. Because, you know, our models are only as good as the data they have. And the data they have is built by humans. And as things change over time, you need new data. And the way we get new data is interacting with the consumer. I think it would be a huge miss for any retailer for any E comm company to lose sight that there are people on the other side of that screen. And that's what you're really trying to resonate with. And no matter how much we can distill this stuff down to get more accurate and we can, right, and that will drive performance that'll lower cost us operate at scale. At the end of the day, we're trying to connect with people. And you can only do that with people power.

Loreal Lynch: Oh, absolutely. Another common use case that we see with respect to visuals is just related to brand compliance. So there's a workflow where we have some customers that are brands and franchises that license, license out their content or their brands, and they have, you know, third parties that will leverage that content. And they might have a whole network of content creators themselves, whether it be contractors or in house creatives that are creating content. And there is a style guide, but oftentimes it's, you know, like a PDF or a PowerPoint and they will give them that style guide. And often it's, you know, 100 plus pages and they might not read something specific in the style guide. And so it has to come back to this brand compliance team that has to review it. And there's like a, you know, an SLA on like a turnaround time. And I think what we're seeing is that just AI it's a very, it's, it's almost like a very simple use case. Not as complex or advanced as some of the ones that we're talking about with respect to the creation of branded content, but just being able to kind of stop something that's off brand in its tracks and reduce the time that it takes to have to go through a human for review. But if AI can just automate and say this is off brand, this doesn't have the right color palette, this doesn't have, you know, this is like a weird, this is against our brand guidelines in some way. We have a capability called visual guidelines that, that's been kind of picking up Steam. Is, is brand compliance anything that is part of the workflows at Wayfair? Is this, is this something that you guys are focused on at all?

Brian Godwin: It's on the horizon for sure, you know, because we have so much user generated content out there and you know, that resonates really well with the consumer, right? When we, you know, depends, depends on the marketing channel. But when you go on TikTok, you want to have really organic types of media that are on there that are, you know, being driven by users or paid users, paid influencers, sometimes unpaid influencers. And you're right, there are brands and style guides that, you know, especially the upper tier folks that are in those networks get to use and there's always a concern that someone's going to go off brand. There's a lot of, like you mentioned with your technology, there's a lot of really interesting research that's utilizing computer vision and, and models that are trained on brands to be able to analyze and detect when something is out of whack and hopefully pull it before it gets to the consumer or at the very least flag it. So you could go to that content creator and talk to them about the brand guidelines. It's early days on that tech, but I think it'll get there. We'll be able to enforce guardrails together. But what I kind of see this going though a little further is as these tools continue to evolve and get commoditized, I think that Genai is going to become just so pervasive in terms of content creation and easy enough to use that even just small content creators are going to be able to use these tools that brands will be able to provide their own models and be able to provide their own actual networks that can be given to influencers and content creators and say, hey, you can use this to create content with your content, amplify it. But it's already trained on our brand standards so you can catch it earlier rather than and later. And I think that's sort of where things are going in terms of like today training models really hard, computationally intensive, takes a lot of effort, trial and error. But I think that's going to keep getting faster, faster and cheaper and Easier where we can literally just take that set of brand guidelines, feed it into a model, and then have something multimodal that can come out the other end. It's not today, but maybe in five years.

Loreal Lynch: Yeah, yeah, yeah. Well, just. Just to close out and to end. I want to ask you, because if that's kind of over the next five years, what's the thing that you're the most excited about more in the short term, let's say over the next year, either in the industry or at Wayfair specifically?

Future of Immersive Content

Brian Godwin: You know, just from, like, personal perspective, Based on my background, I'm really interested in where immersive content is headed both for retail and for entertainment. There's a lot of really complicated problems to solve in terms of creating things like depth in an immersive environment or 3D outputs that we're still struggling to tackle, particularly with gen AI. But even just in the last few weeks, we've seen really great progress going where you can give a single image of something, a product or character or what have you, and get a 3D model out of it. That's like the first step of getting to immersive content. And while VR headsets haven't really caught on yet, AR is starting to catch on more. I really hope that AI can really commodify that type of experience so that people can get to that personalization from a flat screen that shows you what you want to look at to being in that world in and of itself. And I think we're going to get there faster than we think. And I think in the next year, we're going to be shifting from what's on the forefront, which is generative video today, to generative immersive content tomorrow.

Loreal Lynch: An exciting time to be in marketing and creative, I think.

Closing Thoughts

Loreal Lynch: And with that, we're going to go ahead and wrap up. Brian, thank you so much for being with us today. Thank you for these insights.

Brian Godwin: Thanks for having me talk to you. Yeah.

Loreal Lynch: And thank you everyone, for joining.

Brian Godwin: Thanks, everybody.

Loreal Lynch: All right, take care.

Brian Godwin: All right.