Retail Case Study: Unlocking AI to Change Consumer Behavior Toward Profits

With sights set on the digital experience, all retail segments are looking for ways to improve customer engagement, driving a personalized experience similar to that of the in-store experience. To do this, retailers are using machine learning to power visual search and product recommendations to increase customer engagement. Learn how eyewear retailer EyeBuyDirect leveraged visual data and optimization to not only grow consumer engagement, but grow profits as well. Albert Guffanti: Hello, welcome to today’s webinar. My name is Albert Guffanti, I'm the group publisher of RIS and CGT. Today we have a fascinating case study featuring EyeBuyDirect, which will be focused around the topic, “Accelerating Retail Sales Through AI-Powered Visual Search and Personalization.” I'm excited to put a target on how AI is driving sales. We’ve talked about the ways that AI is powering the enterprise in multiple areas, and I’m excited to dig into how AI is driving sales, enabling consumer behavior, understanding and acting on consumer behavior, facilitating consumer demand, and so on. What we're going to cover in this discussion is three areas. First, we’re going to dive into the global trends in retail — what our panelists are seeing in terms of artificial intelligence against the backdrop of what's happening now in the world and looking forward. Then, we're going to dive into the discussion at-hand, which is discussing AI-led digital transformation in the retail industry. We'll get a great perspective from each of our presenters on what they're saying. Once we do that, we're going to dive into the topic at-hand on how AI services are driving revenue. Without further ado, I want to introduce our speakers for today. First we have Alex Alekseev, director of UX at EyeBuyDirect, our resident online retailer. We also have Brendan O'Shaughnessy, the chief commercial officer at ViSenze. We also have Steve Gurney, head of general merchandise retail at AWS. Gentlemen, it's a pleasure to have you on this webinar to talk about this really fascinating topic. I’d like each of you to introduce yourself, your role, and what you do at your companies. Alex Alekseev: Hi everyone, I'm Alex. I've been with EyeBuyDirect for the last eight years. I come from a background of user experience, but mostly I was working a lot with these technologies, as part of user experience activities. In my journey with EyeBuyDirect I've been working on technologies such as AR virtual try-on, for example. Some of the background that we should give to our audience is that EyeBuyDirect sells glasses. It's a challenge for someone to buy glasses online, which is why we have to rely on technologies — that's why I mentioned AR virtual try-on — personalization, recommendation. A lot of it is using AI. Brendan O'Shaughnessy: I'm Brendan O'Shaughnessy, the chief commercial officer at ViSenze. We're a nine-year-old AI company. We were a pioneer in the development of visual AI, and particularly visual AI solutions for retail customers. As the chief commercial officer I run sales, marketing channels, and product development. I'm based in Singapore with a global team. We work closely with many leading retailers around the world in adapting our market-leading AI solutions. Steve Gurney: Hi, everyone, Steve Gurney. I look after general merchandise retailing here at AWS. AWS is the web services/cloud services arm of Amazon retail. We've been born out of retail and naturally, retail is a very large industry for us. Most of my time is working with retailers of all shapes and sizes. Small disruptors right up to the large, established retailers. We help them modernize their tech environment — basically, take it to the cloud and make it cloud native, as well. That gives them a lot of agility to react to customer experience changes and react to things like the pandemic, but also build more efficient optimized businesses as well using cloud services. Guffanti: Clearly we're in good hands with our expert panel today who are knee-deep in the retail market and have tons of experience in leveraging technology. We're going to start off at a very high-level before we drill down into the topic at hand. Steve, I’d like to throw it to you. What are you seeing globally in terms of trends in retail right now as it pertains to this discussion across all the markets? Gurney: It's an interesting area because the pandemic has changed so much. We're aware of the obvious headlines. One of the main things that people talk about from a retailing sense is how digital commerce has exploded. For most retailers, they've seen online digital sales double since pre-pandemic levels. Others, obviously, more than that, but that tends to be the general level sitting right across the retail industry. Equally, we've seen another trend, which is the aspect around retail inspiration. The element that happens before the purchase, before the transaction happens, has moved on digitally as well. That's for not only the online purchase. That's for in-store purchases, as well — all of retail. More of that inspiration is happening online. If we think about the consumers, they're following Instagram feeds, Pinterest pinboards, those sorts of things and the trends that are happening there. New outfits, new inspirations, new trends. All of those things are happening. They're following YouTube influencers, as well. That's a high-growth area in all areas of retail, all areas of consumerism. Things like live-streaming. Although in Alex's area, in APAC, that's really well-established there and in some markets it accounts for 20-30% of digital sales. Those things are growing. If we look at it from a global context, that's a fast growing area in the West. Certainly, in new places like Lat Am as well, they're jumping straight into retailers. That whole digital inspiration is a fast area that we're seeing retailers invest in. This is a good summary of some of those technologies that retailers are investing in. On the left-hand side, visual search. Naturally, for customers, a lot of their time is spent looking at images and visuals on Instagram and Pinterest feeds. If they want to look for an associated product or for available products, then visual search is a powerful tool there. Both in the general sense, plus when you actually get to a retailer's website. There’s a lot of investment there. AR has exploded. There's technical elements and user experience elements behind that. There's new technologies such as web AR, which means you don't need a specialized app anymore, you can do it straight from a browser. Customers want to look at a product, or try-on a pair of sunglasses or glasses in Alex’s case, and they can do it straightaway through the browser. It will jump straight to their camera, and they actually see the products on them or in the room. The middle example uses the Amazon room designer. Homewares is an area we've seen that really take off. I was doing a recent panel with Wayfair and they've seen AR use explode and they expect it to continue. Customers put sofas or lamps in their rooms, in an attempt to see how they work from a color perspective and gaining inspiration there, but also checking things like do they fit? So there's many aspects that go into providing better confidence in making a purchase. That's true whether it’s digitally or actually online. The last area is AI tools. AI is behind the two elements I just spoke about. We're also seeing lots of creation using modern AI tools and machine learning capabilities across the retailers. A good example is Lando creating outfits on the fly so no matter where you browse throughout their website, they're creating personalized offerings and outfits. Things like sizing suggestions, so that you get that product right the first time. They don't have to worry about the customer being disappointed and having to return it, or order multiple sizes to get the right item. All of those things around delighting the customer, delivering a better customer experience, but also optimizing the business. That's a trend we're seeing with AI, it's hitting on both of those areas. As I mentioned earlier, live commerce has been a growth area. There is a lot of activity by retail customers in live streaming. The examples here are from our parent company, Amazon. However, as a consumer, you're seeing retailers do this right across the board — combining maybe presenters plus influencers, and then celebrities in some cases to attract those and make it highly engaging for the consumer. We're seeing new technologies married with live stream commerce. AR and more host interaction is coming into those services, as well. Simple things like chat, but now there's voting panels and you can vote on a new product, which one looks best, etc. Gaming is also coming into it. The whole load of advancements happening in this whole livestream commerce area. We see that actually continuing as well. The last point I wanted to talk about was virtual stores. We’re seeing a lot of our retail customers invest in this area and it's moved onto a dynamic engagement activity now. Whereas before it was kind of like a play thing, building a virtual store and people can look around. Now virtual stores are very interactive – they're linked to the online website, the prices, the live inventory. When a customer enters, they can see what the store looks like, the range of products available, and actually click on items to purchase there and then, or they can go to the store. However, when they’re looking at it, they see that four or five of those items are in-stock, so if they head there later that day, there's a high chance they’ll be able to buy it. All of those things are becoming more engaging. The example here is Ralph Lauren in fashion. We're seeing this across the board in retail. To provide another example in consumer electronics, Best Buy's taken this to another level and is actually staffing those virtual stores with real staff. Customers can have a direct chat and engagement, talk to a blue shirt expert in their case, and actually get buying advice or technical advice on a product, maybe they’ll talk about installation, etc. Again, they’re making it highly engaging. We've seen that in luxury fashion as well, where a lot of the usual engagement that would happen in-store is now happening digitally. Customers can have an instant video call and talk to a fashion expert. It goes back to the point I made earlier, it's about that pre-purchase phase, delivering the inspiration, improving confidence that making a purchase is the right one. Customer satisfaction is extremely high. We're also seeing these interactions serve as feedback channels for retailers. These experiences give them demand signals, ideas about what customers are trending, etc. There's a lot of customer feedback that can come through these new channels as well. O'Shaughnessy: There’s a lot happening in the AI space, as we know. We often get asked by customers what AI technology can do to help them in their business. What are the areas of focus that they should be drilling into and adopting within their own business? We've got a strong global customer base that we've established over the years, which allows us to see a lot of the different ways retailers apply technology – some great, some not so great. We decided to make it easier for customers in the wider market to learn from our experience with the market insight that we can bring to the table. At the beginning of this year, we published our first quarterly index report. It has created a platform to more easily share insights and case studies, providing a statistical view of the growth and adoption of technologies like visual search. It also gives voice to some of the industry's leading commentators and leaders. In our July issue, Lexi Willetts, CEO and co-founder of Little Black Door, a fashion marketplace company, discussed digital differentiation of fashion. She shared her thoughts about how prevalent digitalization is becoming across the industry. Luxury retailers like Fendi are sending DM's. There is an emerging trend of fashion brands switching to direct-to-consumer models with distribution channels via Shopify. It's easy to set up, quick to launch, and then you get the likes of Gucci partnering with Snapchat. There's a lot of evolution going on, coupled with significant growth in marketplaces across the world. This boom in digital commerce results in more choices for consumers. Our instinct is that these days the experience you can deliver to customers, the insights you can collect about customers, and how they shop are becoming increasingly critical. Our AI solutions are largely focused on helping enrich consumer touchpoints, we make it easier for retailers' consumers to find the right product. We're finding more and more retailers looking for help in one of three ways. First, we auto-enrich product attribute data to give an overnight boost to the on-site tech search engine. In addition to helping with SEO outcomes, we enable visual product search, which works really well to complement tech search. This helps match the consumer intent with the right product faster. That works equally well online or in-store. There are several use cases where we work with leading brands in an in-store capacity, as well. That gives a nice cross-channel, complementary experience for consumers and deeper insights for the retailer in question. We also enable retailers to deploy new product-focused recommendation strategies allowing them to easily configure and deploy different AI recommendation algorithms that are delivering revenue outcomes for customers. An optimized computer experience is critical for success. I'm delighted that Alex from EyeBuyDirect has joined this conversation. EyeBuyDirect is a digital-only business, which is growing at 40% year-on-year. We've been collaborating since 2020. They first evaluated how our AI recommendations engine compared with the solutions they had previously deployed. Today, we’re continuing to explore new ways that AI can further enhance business goals and make it easier and faster to deliver the right eyewear to customers. From working with Alex and his team from the very beginning, I know they are very much a data-driven company. Alex, you share some of the consumer trends you were seeing and how technology helps EyeBuyDirect optimize consumer touchpoints to better meet increasing consumer demand? Alekseev: First, I want to say I was listening to you and Steve, and the observation of all the trends from your side is very relatable to us, the business, as a consumer. Starting last year, when COVID started, a lot of customers changed their behavior. Existing customers, potential customers, or existing customers, would go to stores. Now they’re forced to stay online. In the beginning of last year, we saw a huge spike of desktop visitors who are using laptops, for example. Why was this? The overall trends we had previously seen were on mobile devices because more and more people would shop on the way. Then, suddenly, at the beginning of last year, we had a big spike of desktop users. People who are stuck at home, work from home, and are using their desktops. Their experience is going to be very different, and it was quite unexpected for us. We also see a lot of customers that are not used to shopping online – they need guidance to shop online. Before I would walk into a store, talk to a salesperson, and ask, "Hey, can you recommend me glasses?" I don't even know where to start: How do I choose the right lens, the right frame. This website is going to teach me, right? You have 3,000 products, I only need one, help me choose. It's challenging. We have a lot of customers who are new to online shopping. For them, we need to make sure we make it easier. We need to rely a lot on recommendations in order to simplify the process. We need to make things simple and quick, this is where we focus a lot on different technologies. The demand on the virtual try-on increased a lot compared to two years ago. The engagement – as you mentioned, Brendan, we are a data-driven company – we were observing and measuring engagement with different tools of the website. For example, around May of last year, we started to see 20-30% higher engagement with virtual try-on. It was already very high for our website, half of the users were using it, but now about 70% of customers or visitors would actually engage with virtual try-on. A lot of other technologies are more in the background. For example, it might be AI that made certain recommendations that we worked with ViSenze. There are certain recommendations based on the behavior, etc. Generally, it plays a bigger role in our business, in the growth of our business, and in the performance of the overall online part of the business. Guffanti: As you all are talking, my mind is a little bit blown by all the things that are a reality in terms of technology, services, and experiences that we are now able to deliver. It occurs to me that we're in a perfect storm scenario where the technology has advanced just in time to address the tremendous amount of disruption happening in the industry right now. With those two together, we're now able to offer things that were not possible in the past. Let’s talk about this digital transformation that's happening – artificial intelligence and the role that it's playing in driving that transformation in our industry. It’s pretty incredible. Brendan, can you talk a little bit about how much of a lead role AI is playing in this digital transformation that we're seeing in all these areas? O'Shaughnessy: Let me first take a step back. We've been around for nine years, ViSenze was founded in 2012, and it's probably over the last five years that we've seen major brands and retailers begin to explore and adopt AI technologies. Five years ago we were deeply engaged with innovation teams who were curious about what AI was and what it might mean for businesses. Their focus at that time was very much on experimentation. I'm glad to say that much of that experimentation led to production rollouts and growth. Today, we are deeply engaged with P&L teams focused on delivering real business value. Alex just spoke about how data-driven they were. When we engaged EyeBuyDirect in 2020, the assessment was all about data and the impact of the solution we could bring in a very measured way. When our customers look at our AI solutions today, they're no longer curious about what it might do. They are evaluating it based on the impact it can have on their business. That's a fundamental shift in mindset, and that's accelerated over the last 12-18 months. What retailers love about our solutions is the transformation that we are powering in making AI faster and easier to adapt. It's not measured in hours and days, whereas before it was months. We're building tools to allow retailers to configure and optimize AI algorithms directly, so they no longer need data scientist involvement or even a code drop inserted in certain instances. That's a big step that customers love because AI is complex. This technology is very, very complex. One of the key things we're doing is trying to remove the complexity from the customer's spectrum. If you look at Gartner, Robert Hetu prepared a report based on a survey of retailers recently and their expected adoption of AI. Seventy-seven percent of respondents indicated they would have deployed some level of AI within their business by the end of this year. Gartner modeled the assessment of AI value in retail on what they call a “use case prism.” There's 23 use cases that Robert has mapped and researched. I'm not going to go through all of them, but we can certainly touch upon a few of the cases that we believe will be the focus of the industry as we move forward. This prism looks at multiple use cases from the twin lenses of business value and feasibility, measured in both cases as medium to high. The top five use cases are judged to have both higher business value and a higher degree of feasibility today. AI services, such as visual search, product recommendations, and product enrichment all play a role in some form or another when it comes to the use cases identified by Gartner as being at the forefront of retailer's minds. We see that has been quite consistent with the conversations that we have in the market with our customers. When you look at pricing services, shelf analytics, personalization use cases, there are clear leaders in where AI solutions can add business value today. Steve mentioned the rise of influencers and social corners. It's true, brands also have to be plugged into the many channels that consumers are scouring for both integration and brand connection. One of our customers, Yapo, uses a deep sense visual search engine to enable retailers to link consumer and social gallery images with products in their catalogs. Our solution provides greater accuracy than the previous manual effort and saves a lot of effort. The same with fashion, size and fit solutions add strong value to reduce the amount of returns that retailers have to contend with. That's a multi-billion dollars per year business problem that the industry's still trying to solve today. There are also other use cases where we're seen progress today. We're working with a leading fashion marketplace customer to design and build an AI-enabled assortment planning engine that will combine DeepSense AI services with AWS services, like Forecast, to deliver an end-to-end integrated workflow. This will put data insights at the heart of an optimized and semi-automated purchasing process. We're expecting to launch a beta program for this in Q1. There is already a leading global fashion retailer using our visual search engine to power conversational commerce by bringing visual product search to their website chatbot. We expect this to grow over time as more and more retailers see the value of incorporating both visual search and AI recommendations into more of their consumer touchpoints. Finally, I want to touch upon real-time pricing. This is another area that Gartner noted as a lower priority today, but nonetheless one that is of interest to retailers. It’s the use case that many customers are inquiring about. The truth of it is, for some categories, such as furniture, there isn't enough consistency in product data to rely on when you're trying to compare competitive products or products from different competitors, and identify related prices. That's where AI visual search comes to the forefront. It's more accurate to use visual product matching for price comparison than it is for product data alone. We recently ran a trial with one customer who saw a 5X increase in product matching accuracy using our visual product search compared to their product data comparison approach. We're excited about seeing more of these use cases and think that's going to increase in the minds of retailers over the short-term. Beside AI technologies, another element at the heart of all of these use cases is data. Alex spoke about how data-driven EyeBuyDirect is, and they're not alone in that. More and more retailers are realizing the value of understanding the consumer better. That data is critical to understanding consumers, their interests, and their buying habits. We're able to make critical insights available by the type of products that consumers are looking for both within our customers’ own environment as well as the broader market. Product data has become more and more important in terms of unlocking, enabling, or improving many solutions. Frankly, we're excited with what we developed today and our roadmap for developing more data insights that customers and partners will find beneficial to their business. We see it as a combination of those AI technologies for some of the use cases we spoke about. The heart of that is data, and for us that means AI curated, processed, and analyzed product data more specifically. Guffanti: There's so much to consider here. The slide that you showed on the capabilities of Gartner is sort of mind-boggling on all the things that AI can help enable a retailer. It makes me wonder, Alex, as a retailer, how do you prioritize what to tackle first? What challenges to solve? What priority is first and foremost as you look to consider an AI strategy? How did you go about it? Alekseev: You actually hit one of the key topics that we often discuss internally: how to actually prioritize things. There are so many things, especially in e-commerce, where things are happening lightning fast. Generally, we approach things by starting with the goals, what we’re trying to achieve. Whether you're looking for AI to improve internal efficiencies or you want to improve the bottom line, or you want to blow consumer's minds by introducing technology that would be very innovative and be used for PR and media, and create a good buzz. Starting with the goal of what you're trying to achieve and then trying to identify the scale. How big is the coverage going to be? Is it going to impact a small part of the customer experience or can it become one of the core pillars. To give you an example, if the AI or generated technology that you're trying to implement can be used across multiple channels, can be a use for channels, or be part of your store website. If you talk about product recommendations it doesn't necessarily mean that you recommend something on the website. You can also send an email. For example, the introduction of products, you can say “hey, we also recommend these products for you.” This recommendation may be based on products that are similar to what you currently send or similar to what the customer actually browsed or purchased in the past. There is a lot of flexibility there. Then, of course, the impact of it. If it covers a lot of customers or solves a big problem, is it going to be a big deal? Sometimes, yes, we tend to over-hype the AI technologies, but sometimes it's not mature enough – so, maturity and how strong the impact is going to be. The last part is the flexibility. All the businesses are slightly different, so it's getting tricky to acquire a product or technology that works out-of-the-box and delivers maximum value. Usually, you need to tweak it somehow, adapt it to your purpose or your business. Make sure that this technology is allowing you to do this because many things are developed in very specific use cases, very direct use cases, and that becomes a challenge. Especially as you try to grow it. You implement it, integrate the technology and it works okay. Now you're thinking okay, what's the next step? Then you realize that actually, maybe this particular technology doesn't really have space. When we implement a certain solution, we always think short-term, midterm, and long-term how we can use it? How can we expand it? That would be four main parts that would help us with prioritization. Guffanti: The heart of this discussion is really how AI services are driving revenue. Id’ like to get your insight on how you used AI to enable your growth strategy. What kind of challenges did you overcome with AI? What kind of outcomes did you see? Can you take a few minutes to discuss AI and revenue, if you wouldn't mind? Alekseev: The key is to start with a good measurement, first of all. Again, AI is often used as a hype word, everything is called AI. Then, not everything performs equally well. You will definitely need to measure it. In our case, the number of things we use AI for is the product recommendations. To give an example, when we started discussions with ViSenze, we started with the product recommendation and how we should recommend products. Originally, the idea was to recommend products based on certain attributes. Let’s say a person buys round black glasses, then we recommend something similar. However, this approach is not exactly correct because both products might be black, round glasses, but they look significantly different. That's how we started working with the ViSenze AI solution, where we explored two parts: smart recommendation and visual search. The search allowed customers to upload a picture, for example, of a product they liked in order to find something similar. However, given the nature of our business, generally, our customers are not using search that much. So we decided to focus specifically on product recommendations. We implemented similar product recommendations on product pages as the first step. Basically, the experience would be the same: Once you browse the product, you see something interesting. Then, maybe it's not exactly the right fit or the price is not the same and you want to continue browsing. You want to continue exploring more products. We had previously recommended products and now we advance to the next stage where we want to improve relevance. This changed when we started working with ViSenze and switched to products. There are two interesting parts here. First, we've seen immediate change in the customer behavior. The moment we started recommending products differently, we saw that engagement increased. But that didn't mean revenue increased because we might start recommending products that are having a lower margin or are generally less expensive products. This is where I mention flexibility. We wanted to ensure the products we recommend are also aligned with the business strategy. For example, if someone is looking for Ray-Ban glasses, we don't want to recommend some low-priced tier product. This is where we started to see uplift. We improved the relevance of the products, and they started to be much more similar to what the customer sees. In terms of visual similarity, it was much more accurate, it was close to how a person would say, "Yes, it looks the same," or, "Yes, it looks similar." But it was automated because it's impossible to find similar products for the 3,000 SKUs that we have, and then we update them every month. It would be an enormous amount of manual work. Guffanti: You commented earlier about how normally someone would walk into a retailer and there'd be a salesperson guiding them in the right direction. This approximates or gets even better at making product recommendations via a digital experience. Alekseev: Yes, it actually gives, at least in terms of similarity of our products, it’s quite spot on. Then, the moment we were able to adjust the strategy and how we recommend from a business perspective – how expensive and in which category the product should belong. We started to have quite a good uplift. There was a 30% higher click-through rate on this area because of relevance, but there was also a conversion rate uplift on the average order, or like-purchase value increase. Overall, we had an 8% increase for the overall session value. Guffanti: Those are amazing insights. Brendan and Steve, what are you both seeing in regard to product discoverability and smart search recommendations? Are you seeing similar lifts and results across other retailers? Gurney: Any of those areas where you're delivering a more relevant, richer experience. The term hyper-personalization is used a lot when it comes to using AI and machine learning tools to work in this area because they can do a lot more at a higher scale and pick out those niche nuances of the customer base. Then, as Alex mentioned, when a new product launches, you can let customers know that the product will be relevant and highlight it to them, rather than sending a general blast for everyone and hoping it sticks. It’s that personalized messaging that we've seen really take off. Also, social media advertising. Again, analyzing your own customer base to understand those nuances. What resonates with that customer? Then, to the likes of Instagram, "Find the other customers like this and show that product launch advert to them." That’s another area we've seen explode. Not only the personalization of retail properties, in stores, online properties and apps, but also in the social channels when it comes to advertising or email campaigns. O'Shaughnessy: Alex has talked very eloquently about how our recommendation engine delivers great revenue, but we learn from our customers as well. He mentioned what they did on the PDP, but once they went into the recommendation and generated similar products and sought out engagement, they actually had a great idea to create a page that was automatically generated by our recommendation engine. That gave an even greater uplift in terms of conversion per session value and average order value. We listen and learn from our customers as well, it's that collaboration and partnership which is critical. There's a couple of things that Alex spoke about earlier, including prioritization. AI is a complex technology. It's much easier to play with today, but it's a different proposition to create AI solutions that can create real business impacts. Privatization is important when you're seeking a budget or resources to launch new solutions. One of the things we are focused on is trying to remove that complexity for our customers. Earlier this year we launched our retail AI platform, which we call Discovery Suite. It brings together all of our core AI services – visual search, recommendation and enrichment – and links them through a new catalog manager. You can consider that almost like an enabled pin system. When combined with a smart data module, the platform brings quite deep search and recommendation analytics to our customers. With Discovery Suite, there are key benefits to deliver to our customers. One, to make it faster and easier for customers to adopt AI solutions. Shopify merchants today can integrate with these sims in just one hour, which is an indication of where we're going with this. Two, we're giving merchants more configurable control over algorithms, which Alex talked about. His team now has control of those algorithms through our console so they don't need a data scientist or code drop anymore. They can configure and optimize recommendation algorithms based upon the data insights and analytics they're seeing coming through from a platform. Those are the valuable data insights that retail customers are looking for because that's what helps them measure, optimize, and ultimately, generate even more revenue. There's a lot there for customers to work with to improve revenue outcomes through AI. Guffanti: Brendan, can AI solutions help or support catalog enrichment for retailers? O'Shaughnessy: In short, yes. We do it for a lot of customers around the world, and it's also the creation of data. The lack of product data or sufficient attribute data for products is a fundamental problem across the retail industry, it's something we've been tackling for years and something we do very well. Our focus here is on removing the manual process, the inconsistencies, and the inaccuracies that come from people doing what is a very manual, subjective job. The AI engine can auto-enrich a product catalog overnight, it takes a couple of hours depending on the size of the catalog. We work the marketplaces with tens of millions of products and do that for them, which is a great enabler for improving text search, improving SEO, improving granular analysis of products and business, and it's very popular. Guffanti: Alex, how easy or difficult was it for EyeBuyDirect to move from an in-house recommendation solution to an AI solution like ViSenze provides? Alekseev: Moving was relatively easy. Including ViSenze, but also other AI technologies that we were introducing, were relatively easy to start. However, it takes time to maximize the value. This is something to keep in mind, but starting is quite easy. Guffanti: Steve, our last question is for you. How important is data and analytics to this entire conversation? Gurney: Data is highly important. A lot of customers underestimate what they can do with their data today. They embark on lots of big programs to build out data lakes and so forth because they need to do that before getting the gains. We've seen the speed of deploying new generation tools rapidly increase. If you looked at it even two years ago, what you can do now in a few hours would have taken days and weeks a couple of years ago. The point is you don't have to get all your data in one place. You can have it in multiple places and still start using the tools or proof of concept quite quickly (within days, weeks), roll them out, and A/B test them. The ones that work, you roll out and deploy to customers. The ones that don't, you reiterate and either give up, move on, or make them work and deliver that better customer experience or better business optimization. Less so on the analytic side. It’s about using the new generation tools to provide better demand signals. Then, these tools are becoming faster by the day and easier to use as well. That helps everyone use them, not just the experienced companies that have lots of data scientists on-hand. Guffanti: Unfortunately, that is the time that we have today. It went extremely quickly, as interesting conversations typically do. I want to thank everyone for taking the time to listen in on this conversation. First and foremost, I want to thank all of our speakers for sharing your insight. Alex, thank you for providing some of your case study-driven insight as to your journey at EyeBuyDirect. Brendan and Steve, you're doing an amazing job in terms of helping the industry adopt AI and taking advantage of a lot of the transformation that's happening in our industry, I can't wait to see what happens next. Lastly, thanks to the audience for joining us. Have a great day everyone, be safe, be well and we'll talk to you soon.