Hello, I'm James Risley, Research Data Manager and Senior Analyst at DigitalCommerce360, and I'll be your moderator for today's webinar. Thanks for joining us. Today, I'm joined by Imran Choudhary, VP of eCommerce at Crowpeak. On this webinar, we'll be discussing the complexities that internationalizing and localizing eCommerce experiences poses. By adoptinging a thinkglobal, actlocal mindset, retailers can leverage market insights to optimize their international expansion strategies. We'll be taking housekeeping notes before we get going. We'll be taking questions at the end of the presentation, but you can ask them at any time using the chat box on the right side of your screen. You can also use that box to let us know of any technical difficulties that you're having, and someone will help you out with that. With that, let's get started. Imran, thanks for being here. Thank you, James. Thank you very much. So, good morning, good afternoon, good evening, wherever you are in the world. Thank you for joining us today. So, what are we here to talk about? As I mentioned, we're talking about a topic of how do you think global, but act local? And we're going to dig into a little bit as to what that means. But fundamentally, how can we localize the eCommerce experience to drive higher revenue, growth, success, and engagement per region? Okay. And what are we going to talk about in terms of how we talk to that topic today? Well, CrowdPeak, our eCommerce business, we focus in on product discovery. So, how someone discovers a product, say through search, how they go through the shopping experience with regards to merchandising, and how do they then convert maybe with higher average order values through things like cross-sell and upsell. And the insights that we're going to share today are from two areas. So, the one is an independent study that we did into the market and the state of product discovery as we see it globally. And then the other part is some interesting experience and some best practice of the work that we've done with some of our customers. So, in terms of our customer base, the good proportion of them are international. So, we've got some best practice and some guidance of what the leaders are doing in this space, which hopefully can inform and advise you, especially as you think to your own, in terms of your own strategies and what you are looking to do there. And then in talking about that, as CrowdPeak, we power over a third of the top 25 UK retailers. And we also power a quarter of the top 50 European retailers. So, we hopefully, what with that understanding of our customer base and how we power those businesses, but also the industry insights we're going to share today, hopefully you feel a little bit of credibility in the things that we're going to share with you today. So, let's make a start. So, this is today's agenda. We're going to go through the challenges that we see when organizations are trying to go international. We're going to look at what leaders are doing with regards to localizing their merchandising efforts. And the reason why you can see search as a second part, as it were, in the agenda is because fundamentally in the market, we're not talking enough about merchandising. As a market, we seem to be very hyped and very focused around search. And don't get me wrong, search is extremely important and a core capability for us. But where we are seeing kind of the real success is with merchandising. And that's what we're going to focus the majority of today's session on. There'll be two takeaway references that I'm going to touch on, and then some final thoughts and recommendations. So, let's make a start. So, this research that I spoke of, so we commissioned this independent research, about 300 plus merchants, so independent research done by London Business Research House. And what we found was 87% of those merchandisers highlighted product discovery. So, those search, merchandising and recommendations, there's both an opportunity, but also one of the biggest challenges that they faced with regards to going international. So, a lot of the insights today, come from that independent research. And when we think about going global and acting local, so trying to provide that localised experience across multiple regions, there's typically three challenges that the report kind of identified. As you would expect, language. Language is a very big one. Everyone talks in a different language when you're starting to go international. So, how do you deal with that multilingual requirement? But also, culturally, there are cultural contexts around language that we need to be mindful of, and it goes beyond just synonyms and variations. But you also have things like product attribution. So, that could be around certain product tagging, certain product attributes that may vary from one region to another, and how do you go and deal with that challenge? So, that's one of the big challenges that we see. How do we tackle language? The other one is preferences. And this is where a lot of the leaders, the retail leaders are spending a lot of their time. A lot of the language piece, we're going to cover a little bit as part of the search conversation today, but a lot of that language piece can be dealt with with AI. But it's the preferences, whether if you can understand the buying preferences and priorities per region, that's where you can really drive better merchandising experience and better performance. And that's beyond just the different seasons. So, for example, in one region, price might be a really big priority, whereas in other regions, it may not be where it's maybe a more wealthy region. So, we're going to talk a little bit about that today. So, when we think about preferences, how do we kind of understand that local market beyond just the season? Is it summer or is it winter? The usual California versus, let's say, Moscow kind of differences in terms of kind of temperature. But how do we understand the buying preferences of the buyer in that region and change and assort our merchandising to accommodate and get more success per region? We're going to touch on that today. And then the third one, and this is one which probably isn't spent enough time on, but it is one that came out of the study, is how do we manage our teams? How do we scale our e-commerce teams to be able to trade successfully in all these different regions where each region may require a different strategy? So, how do you manage them? How do you deal with the access controls there? How do you kind of ensure that they're prioritized in their time in the correct way per region? But also, how do you then drive personalization? And as you would expect, AI comes into some of that conversation as well. So, when we did this study, as I say, about 300-odd merchants, these were the three big factors that came out. Language, how do you deal with these multiple different languages? Preferences, how do you really personalize the shopping behavior and your merchandising to drive success and engagement per region? And then third, scale. How do you scale our teams? How do we scale our merchandising capabilities when we all know that resource is always a challenge? Now, what was quite interesting was that out of this study, just under 50% of those that surveyed, they felt that their tech stack could not meet the needs of internationalization. So, this is a really big challenge that we are looking to address with our solutions. So, how do leaders localize their merchandising effort? Where do they spend the most of their time? So, what we found from our own customer base and this research is that a huge proportion of time is spent on controlling the merchandising. So, that's the assortment of your product listing pages, the assortment of your carousels. How do we optimize the merchandising per region against your different goals? Now, virtual merchandising is really starting to become a trend, especially in fashion, beauty, and luxury, where you're trying to curate an experience and maybe even replicate the experience that you might have in store into one that's online and aids with that discovery and tells a brand story. So, visual merchandising is really quite quickly actually becoming a big priority for a lot of leaders. Personalization, you would expect personalization through AI to drive that conversion and those average order values. You would kind of expect that that would be where they're going to be focusing their time, especially as they look to automate. So, how do they automate and optimize to deal with that scale problem that we spoke of a moment ago? So, one of the interesting things that came from a study and what we see in our own customer base is the strategy that these retailers and these leaders are implementing. So, if you think of a product listing page, it might be a category page, maybe a search result, but you've got that listing page, the big matrix, as it were, of all your products. Where the leaders are focusing their time is they're not assorting by, say, bestseller or just purely conversion rate or number of clicks. They're taking multiple data points and they're bringing all of those data points together into some form of business rule and they're weighting them. And by bringing in different data points and then weighting them based on your priorities, you can then start to be really quite crafty and take a crafting approach in terms of a strategy that works per region. So, let's give it an example. So, let's say you're a fashion brand and you've just launched a new collection. So, you might take this new in strategy. And this is one that's quite common, especially with our fashion and luxury brands. So, you're taking the data point of the newest products that have hit your catalog and what you're starting to see is the highest views, the highest purchases, but equally, very importantly, especially with regards to demand as a whole, the supply in luxury, the stock availability. Because let's face it, if you might want to show new in, but if you haven't got a spread of sizes or the stock availability is already running low, they may not be the best products to be promoting at the top of your kind of listing pages. And these are the kind of things that we work with our retailers to build these strategies. These are generally kind of the five big ones and most common ones we see. But how can we pull in various data, maybe even from third parties and create what we call a ranking cocktail, a cocktail of these different data points that can then create a rule that then drives some form of assortment. And those that are doing really well with this is taking these rules and then applying it to visual merchandise. And I'm going to show you a little bit of that in a moment. For those that get it right, you can see a dramatic increase in your PLP to PDP. So, on average, we see 22% improvement from going from the product listing page to your product page by taking a more, let's say, strategic approach to your merchandise and strategies. And you can do this per region. And we'll show you some examples of that in a moment. Now, by applying these kind of rules and then leveraging this data and automation, you can then also reduce the dependency and the time on mundane tasks that your merchandise would be taken by doing this per region. And on average, we see a 70% improvement in terms of time efficiencies in building these kind of capabilities in your merchandising and then deploying them globally and adapting them per region. So, let's have a look at some examples of this. So, here we've got a fashion retailer. And what they've got is a product listing page. And they're wanting to blend a little bit of all their different categories. And you could say that it's a bit of a very large shop to look. So, this might be the edit category or it might be the new in. And what we're wanting to do is apply some visual merchandising. So, you can see they've done that with kind of the tops at the top, pardon the pun, the bottoms, and then also the shoes. So, you've kind of got, just in this example, a very basic rudimental example, visual merchandising. So, where you're sorting the product listing page to try and promote that discovery. Now, each of those rows, you can apply what we call ranking cocktails. So, these are these merchandising strategies that I spoke of. So, here at the top, what you might have is the newest in, the number of views, and then the number of purchases. And you're blending those into a rule to define that row and how that row is assorted based on local data. And what you can see here is the difference between two regions. So, Australia, let's say it's very hot right now. UK, I can tell you, it's very dreary. It's very rainy right now. You definitely want your jacket. But at a rule level, it kind of stays the same. But you're using the local data to inform what you're actually going to surface as the product. So, here, just as an example, it's t-shirts, whereas in the UK, it's going to be jackets. A very basic example. But again, Australia, you're kind of wanting the bottoms. Let's go for shorts versus trousers. And then sneakers, it's kind of the same. But what you can see here in Australia is the taste is very different to, let's say, the UK. But that kind of global rule set that you've deployed globally across all these different regions, if that's the right strategy per row or for the complete kind of listing page, you could define it, deploy it, and then use that local data to dynamically change what products you surface based on those local preferences. So, this is one of our customers, Adidas, who do it extremely well. So, you can imagine Adidas, how many countries they're operating in. And what you can see here is the UK and then the German site. So, the products pages are kind of the same in terms of women's shoes, kind of what's it new in. And you can see the merchandise and strategies being actually globally applied. But again, you can see local data feeding the preferences and therefore what products are being surfaced. So, for example, the bold pink shoes, they're still on the same page. They're just in a different assortment. And that assortment is based on the local preferences that you've identified and pulled from your data. So, by applying these global rules but leveraging this local data, you can see how the assortment changes and it has a higher engagement rate based on those local preferences. Now, what's a dynamically different example is Hermes. So, Hermes in the Netherlands, it's more of a casual approach versus France, which is more chic. Again, you're seeing this visual merchandising element coming from the local data. So, you can see that the local data is being used to provide personalization and that's a regional level. And of course, AI is an element of this. So, we see a lot of our customers blend in those kind of rules and those data sources and then leveraging AI to provide one-to-one personalization. And you could potentially do it across the whole listing page or you could actually just split out one of the rows. Or number of rows on that listing page where you're kind of peppering this personalization. And that could be based on various trends, various clicks that the users have, various preferences they've demonstrated previously, even various marketing or even in-store purchases they've purchased, in-store purchases they've made. And you can feed that data, the products that they purchased and where they purchased at what value. And all that information can come in to drive this level of personalization. And then the last thing that I want to talk about here at Crown Peak is about this balancing act. This balancing act of curating and using the merchandise's tacit knowledge and then AI where you're starting to use algorithms to then generate personalization and blend in these two things together. Now, when we start to talk about personalization, you kind of need to start using AI. The volume, the data, and to be able to do a scale, you need to be able to leverage this AI. And that could be just the listing pages as we've seen already. It could be, let's say, the cross-sell here that you see on the left. Or it could be, as many of our customers do, a full shop to look. And how do we do that? Well, first of all, there's that data collection. So we can be fed by all data, even third party. And even in-store data is very common for us. And we collect that data. There's then a level of enrichment. So very often, we find that product catalogs, the product data is quite thin or it's not full. And we need to do a level of enrichment there. And often, that enrichment includes localization based on different languages or different terms that are known in one region. And then we use that and build it into what we call our algorithm orchestrator. So we are in an industry right now, but we are trying to challenge Invert. There's a lot of vendors in the market that talk about their algorithm and they are AI can fix all use cases. And that's not right. We know from our own experience that you pick the right algorithm depending on the different use case. There's a common example that we saw with C-Discount, another one of our customers, where they identified opportunities to use one algorithm for new customers versus a very different algorithm for returning customers. But depending on the use case, be it the listing page, a recommendation carousel, a checkout kind of upsell or cross-sell, you pick the right algorithm. And to be able to do that, you need some level of A-B testing, but also a platform that has an algorithm library that you can pick, choose, and optimize and blend with, or even a platform that allows third parties in. So we're seeing a rise in terms of bring your own algorithm. And that's something that we encourage because we believe that this ecosystem needs to be more open. And it needs to be more open so that you as merchandise and we as a vendor can continually optimize and find the best algorithm combination or model to be able to deploy to a certain use case. And of course, there needs to be business rules. So what we find very often, especially with black box approaches, is that the algorithm will pump out a recommendation, but there's no level of control around the brand story that you're wanting to tell, be it, let's say, if you're fashion, beauty or luxury brand. And especially in luxury, it's quite important. So I'll give you an example. So if you used a algorithm or a platform that was black box, no controls, and it was looking at the data that you are feeding it, it's very common that you could see, let's say, a shopper buying Dior, also being interested in Chanel. Okay. There is a linkage there. Someone who buys Dior very much probably likes Chanel, and is also interested in Chanel. And the data is informing the algorithm that that makes a good idea to, let's say, have a recommendation carousel where you've got Dior and Chanel next door to each other. But if you work in luxury, you know that that's a bit of a no-no. From a brand guidelines, a brand story, you actually don't have Dior and Chanel alongside or adjacent to one another. You kind of keep them separate as different brand stories. But the AI in a black box scenario doesn't know that. And that's why it's very important to have this business rule element where you as a merchandiser can apply a rule against the AI so that you have these exceptions or you have this control whereby you're given the guide rails to AI to ensure that the brand story is told correctly and to how you want it to be done, regardless of the data that's coming through. So for us, it's all about co-piloting with AI versus the primary driver being AI, if that makes sense. Okay. So the other element I wanted to talk about was control and management. So you can see where we're going with the agenda. You talked about the challenges that we're seeing in the market. We talked about what the leaders are doing. And now we're here in terms of control and management. So if you imagine, as with many of our customers, you've got maybe 30, 40, 50 different regions, and you've got different personnel, so merchandisers in your team, managing different regions, or even different regions at different category levels, depending on how big you are. If you're using traditional, let's say, product discovery or search or merchandising within your e-commerce platform, let's say like Salesforce, which we do actually do a lot of work with, it's quite limited. So you will find that you can't inherit or copy rules. It's a case of you having to build them individually per region and then manage them. It's a very big time soaker, as it were, because there's a lot of manual tasks and a lot of those manual tasks are a burden. And then continually managing them is quite difficult because you're having to do it for every different region, for every different category. So for us, what we've done is we've built a management layer where you can define the different regions, the different categories, and take those merchandising rules or strategies, and then they can be inherited across all the different regions that you require. You might want to tweak one or two regions, but then what you're what you can do is you can leverage that local data to dynamically change what products you're actually surfacing based on that localization. But the ability to inherit and the automation of that rule base really helps a lot of our merchandisers ensure consistency, as well as control of that consistency across all these regions, let alone the execution of them. Okay. Ah, perfect timing. I forgot about that click. So on our average 60% of our customers find that they see an improvement, so 60% improvement into their merchandising. Especially when going international, based on this unique capability that we have with regards to managing categories and sites in this international way. merchandisers control within their e-commerce platform what they can do what they can't do what they have access to and what they don't and this is really important especially for our international teams that have a large large number of merchandisers that are operating in very different ways so here you might have let's say Michelle who has global kind of parent view over everything then you may have Tanya who's just looking at kind of the skincare within France you've got maybe Jess who's just purely looking at one element of the category of that skincare and they're sharing these global rules so they're sharing these global rules they've got access to their area that they're managing and trading and therefore can make their individual changes if they need to their tweaks their optimizations just purely at their individual level as to where they need to without kind of affecting everyone else but they've benefited at least from that global rules being deployed to all of those regions and all of those categories this is really really critical for those international brands and it's an area that really helps drive that efficiency and that focus okay so let's talk about search so we talked about merchandising we talked a little bit about personalization and AI talked about control and management so that was the scale piece so let's talk a little bit about search there's so much around the search topic right now it's really just hygiene okay but there's an area that I want to talk on especially when we start to look at international and where the language piece comes through and an interesting trend that's coming about so the first reason why is why is search something we still need to be keeping kind of a thumb on the button with and making sure that it's a tip-top condition well again from the survey that we did we saw the high truth rate as it were but you know consumers do shop from those retailers where the quality of a search result is high so 68 kind of weirdly very close to that is where we started to ask about what about complex search queries so the complex search query could be let's say four words or or more so this is where the semantics of search start to come in understanding intent and that's quite difficult for traditional search and that's where you kind of need these semantic search capabilities and AI to really understand that intent and ensure the quality and the accuracy of a search result and that's the bit that I do ask you to focus on today because a lot of the search platforms that we see especially within the out-of-the-box e-commerce platform they're very manual in how they operate and therefore again requires a lot of time and a lot of manual effort to continually optimize so I'm going to explain where they kind of generally fit and where we see an opportunity an opportunity to drive improvement around engagement accuracy but fundamentally revenue okay so this is a generic search term that we often see so you've got the volume and the frequency of a certain specific search query so as you would expect really high volume with say a one-word phrase such as shoes and the probability of conversion is quite low someone's that they're really you know they start a bad discovery or browsing journey they don't really know what they want yet they might not even decided to purchase it they might just be socially and so one word phrases high volume but low probability in that intent to actually convert now we get to the kind of the second two word phrase where you might have that same men's shoes and the probability slightly improves the volume's still really high the probability slightly improves but again they're still kind of in discovery and then you get to the real complex piece so there was a recent piece of work that Google kind of launched around the number of even keywords they're seeing in search queries has dramatically improved as customers and consumers have a higher expectation in terms of being able to drive results that are accurate and relevant to very complex search queries. And we're now seeing it very much in the world of e-commerce. So here you may have three word plus. So red Adidas men's running shoes. I know exactly what I'm looking for. I'm kind of in this probability of I am because I know exactly what I'm looking for. Show me a really good result. Show me a really good recommendation. I'm most likely probably in a good place to probably go and convert and purchase them. So this is kind of where the money is now, where most teams spend their time, especially with traditionally come off. They spend their time manually optimizing here in the one word and the two words. And it kind of makes sense because that's where the highest volume of the queries are and where the highest volume of the queries are. That's probably where you spend most of your time where they don't spend time. And it equally makes sense is the complex queries because the complex queries, the variation of the query is near infinite. It's huge. And the volume of each of those different variants is really low. So it just does not make sense for teams to go and focus their time there. But that's the thing. You should. You should. Because that's where the highest probability of conversion actually is. And this is where we see AI fit. So regardless of the type of query, how we can leverage this with different languages, as well as deal with this complexity is where I can really help. So on average, what we see when new customers come to us is 20% of all their search queries fall into what we call this long tail. And this is what we can leverage AI with to automate and optimize here so that you can start to reduce what we call that zero search result or really inaccurate result to nearly zero. The idea is to get rid of this long tail by leveraging AI. Then what you can do is start to leverage AI with regards to cinnamon creation and optimization with regards to the one and the two words and then leverage some of those merchandising strategies in terms of how you surface the presentation of that result, as we spoke of earlier today. But if you're not leveraging AI to tackle this long tail, I do encourage you to go and look at it. Go and pull up your Google Analytics and look at the percentage rate of your zero search results and see if you are close to that average of 20%. Because if you are, that is a revenue opportunity, a missed opportunity right now. But you could go and capture leveraging technology like ours with regards to AI to go in and remove that zero search result. And get some of those results that are relevant and hopefully get some of that conversion for you. And the multilingual piece starts to come up here. So when we start to use large language models and across multiple languages, we can start to actually look at it as a whole. So here what we see is Mango, a client of ours, and we're starting to see how they've managed their search results, leveraging AI to automate, but some new trends that come in. So we all think of, okay, well, if we're in France, we're speaking French. If we're in Spain, we're speaking Spanish. But actually we're starting to see that even evolve. So as a society, we're all becoming quite bilingual. There are certain words that we feel more preference to use in a certain sentence, one language versus another. And we started to see it with Mango. So here you've got Vestido with floral patterns. So for those who don't speak Spanish, I will say I'm not one of them, but I do know Vestido is dress. So dress with floral pattern. So if you imagine a traditional search engine understanding two different languages in one search query, let alone a complex query, such as Vestido with floral pattern, that's where AI really needs to come in. And that's what we are starting to see as an emerging trend where your large language model and the AI search behind what you're powering in terms of that search bar needs to understand not just the local language, but the other languages because they're starting to come together and becoming bilingual in the search queries that users are actually using. So that's one to look out for and ensure the platform of your choice can manage those kinds of things like ours. Okay, so two takeaway references. These are available via our site, but Harvey Nichols, a very well-known reputable brand, they went international with us and they wanted to replicate that in-store luxury experience online. So visual merchandising was a very big part of their work with Crown Peak. What they also wanted to do was automate as much as they could with their international site. So as I say, deploy global rules, but then adapt them to local data because as you would expect with Harvey Nichols, a good proportion of their customers aren't just based in the UK or Europe, but even the far east. And there are different buying preferences as well as trends, and there are different ways that they can adapt to the technology that they need to adapt for that they've been able to achieve with our technology. The AI search element. So previously their search was finding it very difficult to actually give a decent search result and an accurate one. So they had a huge amount of redirects so that they just didn't get a 404 kind of error page or completely zero result, but it was still disjointed. Yes, there was a redirect to something, but it still wasn't relevant. As they deployed our AI search, they were able to automate their search results and optimize them so that they were relevant and they were accurate. And therefore they pretty much reduced to near enough removed any redirects going forward. So for the buyer or the shopper, the consumer, they were able to get what they were looking for straight off. And for Harvey Nichols, they were able to remove the manual burden of search as they allowed AI to be able to do this. That as well as how the visual merchandising and the merchandising being localized also enabled them to drive higher engagement and ultimately a 1% increase on their overall online revenue. So, you know, if you imagine their total online revenue, quite big, chunky numbers. The other one is LaraDoot. And this for me really shows kind of where AI fits in terms of personalization. And it's a little bit different because a good proportion of this use case is actually around their marketing off the website. So their email marketing based on what they saw online and also in store. So the AI component, and imagine this challenge. So LaraDoot, home fashion retailer, they've got over half a million products. They've got a customer base of about 10 million plus across 20 different languages. So if you imagine, let's say, just even the email marketing, ensuring that they're relevant in each of those different countries based on the language, based on the preferences, based on that customer's behavior across 10 million odd customers is really difficult. So that's where they've leveraged our AI, not only on their site with regards to search and merchandising, but they were able to push it out. So they had that consistency both on site and off site with their email marketing to ensure true one-to-one personalization. And we saw a huge improvement to click-through rates based on the recommendations and just overall success in joining up both online and offline with our solution. So some final thoughts and recommendations. I would say if you are going international, yes, search is important, but you need to go beyond just the language around search, but also look at how you're implementing merchandising based truly regional and local and can meet the shopper in their region, in their country and to their preferences. Okay, so that's hopefully something that I've showed with you today, not only why it's important to provide that differentiated experience based on those local preferences, but also the importance of looking at global rules and automation to deal with the scaling element. Visual merchandising, especially in fashion, beauty and luxury, this is the secret sauce to really differentiate, aid that discovery, create engagement and that differentiated experience. But applying those merchandising strategies behind that visual merchandising is where you ensure success and profitability. And then AI, please, there is no silver bullet with regards to the AI algorithm world. There is no silver bullet. One AI model can do all use cases. We see in our market lots of people talking that their AI is the best. The fundamentally we need to drive a more open ecosystem with AI where all algorithms can be benchmarked against one another and you pick the right model for the right use case. So please, when you're looking at any of these AI technologies, you want to be asking about the transparency. So why is it making that recommendation? You want to be understanding, can they allow for bring your own algorithm? It might be a third party. It might be when your data scientists have kind of created. Can you bring the algorithm into that platform like they can with ours and benchmark them against one another or even blend them? OK, don't get stuck with a black box or walled garden where you can only use that vendor's AI and nothing else because it's going to limit your kind of ability to optimize and drive that success. And it is a case of optimize, optimize, optimize. So look at AI as a co-pilot rather than purely the driver and just be mindful of those walled gardens that we're starting to see in our market. OK, James, thank you very much. Thank you for your time. James, I'm not too sure. Have we got any Q&A that you'd like to make? Yeah, I was hoping we could pick your brain for a few extra questions here before we wrap up today. A lot of great insight there. So I just want to get as much out as we can out of this webinar. So you talked a lot about, you know, if you're internationalizing, how are you looking at all these new metrics to come in? But are there a few core metrics that retailers should focus on to measure their success of their internationalization efforts? So it's not really, I'm going to say it's not rocket science and that sounds really bad and I don't mean it to be disrespectful, but, you know, it's just the basics. You know, it's the click through rates. It's the page views. It's been able to capture that data at a local area, a local kind of regional and then leveraging that data back into your global wall. So there is no, I'm afraid, secret source in any, you know, certain ratio or certain metric I can share. Well, that's good. It sounds like the retailers who are already comfortable with those metrics locally can feel comfortable then internationally. Using similar metrics. Correct. And what we often see is folks don't know the region's preferences. So we had a situation, I think it was with Puma, where at the start they were kind of testing different strategies with the data that they were being able to pull. But what we were able to find was, let's say in Switzerland, you know, prices are an issue in Switzerland. People aren't sensitive about prices in Switzerland. So therefore, actually full price, full margin, top of your listing page, they probably like that. Whereas in the UK, you know, in the UK, we love a sale. We love a bargains and price is sensitive. So maybe the assortment changes based on price, whereby maybe you put the lower price items or a medium prices at the top versus the full fat, just launched high price kind of list price. So this is kind of where we see that the metrics don't really, there is no kind of new metric, but it's more a case of deep diving into understanding that local data, but having that rule or that strategy to cater for them very differently. Does that make sense? Right. Rather than kind of using the metrics to verify your own assumptions, you have to look into the metrics and let them lead you a bit more. I imagine we are going internationally. And again, optimize A-B tests, you know, constantly A-B tests. A lot of our customers, they have kind of these global rules, as I mentioned. They deploy them as they are in a new region that they may be testing. And then they start to play with them as they start to learn a little bit more about that local, that local kind of preference, as it were. Yeah. Right. So that kind of the next question here, how should companies manage the balance between these global brand consistency that they're trying to keep? Whereas they may also want to take some localized merchandising strategies. How do they strike that balance? If it's a new region, go with your global rules, but you know, generally work and then go and start to test them. There are certain things that you definitely want to do. So in certain regions, as an example, you don't want to sell at a certain time versus another. You know, you want to go fix your data in terms of your product attributes. So, you know, when you're first starting, just let's say price, price of different variants, different regions. So again, those are like some of the hygiene kind of factors that we work with customers on as they're moving into a new region. Let's get the hygiene done right. Get you live, start looking at the data and then start adapting accordingly. Yeah. Great. And so what are some of the key challenges that retailers face when trying to implement the AI driven personalization at larger scales when they're turning to a global audience? Don't believe the hype. I know, I know that sounds really bad because I'm a vendor. Yeah. And we talk about AI and all the great stuff. AI can do it. It can do a lot of stuff, but unfortunately, you know, we read too much into the expectations and set the expectations very high. Okay. So one, be mindful of your expectations. I think the second one is actually education. A lot of merchandisers in certain verticals, they don't want to let go of their manual control. Especially in luxury, they know how they want an edit or a curation to look or collection to look. And 100%, they have that tacit knowledge. Okay. But underneath that tacit knowledge, that's where AI could actually help them, especially if you're thinking multiple regions, you know, you're talking 50 or 100 different regions that you're trying to do this for. Yeah. AI can help take away some of the BAU tasks, but it requires a little bit of them stepping into the world of AI. And I mean, no disrespect, but AI needs some education and it takes a different, it is a bit of a leap to start looking into large language models, different algorithms, testing different algorithms, starting to look into the data and how it kind of works. It does require education. And it's something that we help our customers with. And we actually co-kind of work with a lot of our customers' data science teams, to try and improve the awareness and the education to merchandisers, just to try and aid them on this journey of really understanding AI, where it fits really well, where it doesn't, where it can help them in their job, where it can be a challenge and actually could be detrimental. So the Dior Chanel example is a classic one, especially in luxury. But if they can get it right and they can start to play with it, test the water. So let's say the sales category, that's always a very common one. The sales category, everyone just wants to get rid of that stuff. Yeah. So that's an easy one to go dip your toe into the water in terms of leveraging AI. And what we've actually started to see is the more they learn, the more they play, the more that they test, the more that they actually kind of let go underneath the guardrails that they've given. So I think one of the big things is kind of education and just being open to test these things. Yeah. They seem so magical at the beginning, but I do feel like you need to fully understand them before you fully give yourself over to them. And probably, like you said, I like that. They're the co-pilot. They're not the driver. They stick in that role. I think it's a really great way of thinking about it. Yeah. That's where we think. That blend, as I say, the manual curation, that tacit knowledge, and then the leverage in AI to take away the burden, that kind of blend, the co-pilot. Yeah. Yeah. It's a great analogy. I really appreciate that. And I thank you for all the insights you've given us today and all of our audience for listening. We're running out of time here, but that was a great, great informative session for us. There is going to be replays of this webinar available. So if you do want to share this around. Also, after we conclude, you'll see a pop-up window on your screen asking you to provide feedback on today's session. If you can, just take a few minutes to fill that out. It helps us plan future webinar topics and understand the presentations and how they work. So just once again, I want to thank Imran and the whole Crowdfeek team for helping us out here today. It was a great session. Thank you, James. Thank you all. All the best.