Name: Webinar Mixpanel Final
Uploaded: 2026-06-25T15:49:18.190Z
Duration: 44 min 56 s
Description: Webinar Mixpanel Final

Transcript for "Webinar Mixpanel Final": Everyone, welcome. Today, we're going to be talking about the new rules of ecommerce and discussing product strategy with, Artan and GK from Sun and Midjuri. Today, we'll we'll have a, hopefully, a lively discussion to answer a lot of questions, live in the chat from the audience. But, you know, there's a couple of things that have been top of mind for us. I think we can all agree b to c product experiences right now are very much at an inflection point. If you think about what happened when everything shifted from web to mobile, every single team out there, RB builder had to rethink discovery, navigation, checkout, pretty much the entire experience. The companies that move fast won, and everyone else played catch up. I'm Giovanni Brydon. I run our global solutions engineering team here at Mixpanel. And so I get to help our customers think about what it means to stay on top of their product experience. What's working, what's not working, and how they can all move fast enough to matter. Because we're in one of those moments again, except AI is making it move faster and it's harder to get your arms around, just how the landscape is shifting. AI is changing how customers find products, how they evaluate them. It's also changing how they expect to be talked to. Right? That that shopper expectation is so, so, so much different. The personalization bar absolutely keeps moving and the experiment cycles keep compressing. You might ask yourselves, like, where is all of this potentially heading? Agenca Commerce, AI doing your shopping for you entirely. Analysts are actually projecting that it can handle up to a trillion dollars in transactions by 2030, and that's in The US alone, which represents a quarter of all ecommerce. There's even a website called rentahuman.ai, where agents can hire humans for tasks that they can't handle themselves. So what do you actually do with that as a product team? That's what we're here to figure out, and I've got two people with me who are living it. GK from Imagery and Artin from Sun, a d two c jewelry brand and an AI native start up. So, Artin, I'd love to get a quick intro from you, and GK, and then we'll get into the meat of the discussion. Beautiful. Thank you, Giovanni. Yes. My name is Artem. I'm a founder and CEO of company Sun. Sun is the platform, of the AI generated podcast, audiobooks, and audio courses that, our users can generate on demand. They, and all our content is interactive and hyper personalized. Yeah. I'm based in, Sunnyvale, California, and, happy to be here today. Excellent. Good to have you. Hi, everyone. I'm GK, based out of Toronto, Canada. And I lead the design teams at Missouri. And for those who don't know, Missouri, it's a b two c jewelry brand, omnichannel retail brand. So we have more than 65 stores across US, Canada, UK, and Australia, and and also a significant portion of our businesses is online. And I lead the digital teams here spanning across digital product for web and app. That's your traditional ecom teams, site merchandising, business operations, product teams, and engineering teams. Pleasure to be here. Excellent, GK. Thank you so much. So we're gonna go ahead and jump into some questions. And, again, folks in the audience, as you have any questions of your own, as you have any reflections, go ahead and type into the chat. We'll try to leave some time at the end to answer any questions that you have for these folks as well. But first, I think one thing that we can all agree on is that the shopping experience right now is very much changing. Just like around 2014 when everything really pivoted to mobile, there are now very new ways of discovering products, of interacting with your brand, and, experiencing, you know, a site, a mobile experience, what have you. I'm curious. How are each of you seeing your customers using those new AI interfaces, and what's changing in that browsing behavior, and how is it different? Artin, I'd love to hear from you first. Yeah. That's a good question. You know, my whole company is built around the new consumer behavioral patterns. Basically, at Sun, we don't, promote the browsing experience, as the default option. And instead of brow instead of searching for what exists, you actually generate exactly what you want. So in a sense, it's actually, feels like, with AI technology, the experience becomes more human. Because before, you will just click, you know, like, it's just a silent clicking. And now you're actually using the natural language to navigate to in our case, digital content, to navigate to the content you want. You just ask, in the natural language and the system behind finds, the way to deliver you, whether it's a implicit, whether it's a explicit request when you actually say what you want or the implicit that derives from the your profiling. But still, the interface are more natural language driven. And, of course, the, as you mentioned previously, like, the degree of personalization is, also completely new order of magnitude that it was before. Because, what we've seen that, like, initially, what I would expect that my customers is to just have, like, a single prompt. Like, I want to learn about the history of, I don't know, Croatia. But, what ended up being is people actually, injecting whole paragraphs with the very rich context. And that, interaction brings so much more data, so much more context about my users, which allows me to provide a much richer and better experience across the whole platform. So, more context, more natural, inquiries, more personalization. That's what, we've seen. That's amazing. GK, I'm curious how that's showing up right now for Majuri. Yeah. There's definitely a lot of flavors from what our team mentioned about, his side of the business. Although for Majuri for being more of a d two c, online shopping brand, we're seeing it slightly differently as well where unlike traditional paid media, social media marketing where you would see the ad, click on it, and come on it. Now people are interacting with their favorite LLM model, be it GPD, Gemini, or Claude, or or any other. And a lot of discovery is happening in those models, where people are asking questions about this is an occasion that they want to buy a jewelry piece on or they want to buy a gift for someone. And so they're asking for ideas. And that's where the discovery starts rather than a Google search or or or an ad on Instagram. And they're having these conversations with their, favorite AI agents, which is starting from, I want a gift I want to buy a gift for my friend under $300. And then that gives them a list, and that that list gets shortened down with different filters, with different variables. And that's where the power of content comes into picture is the better content that we have on the site, the there's more chances of, our our products being surfaced on those AI agents. And the other piece is how people are talking about your brand in the different social medias, like Reddit, and more substack and more community based forums. And because all of that content is also referenced by all of these LLL models to to surface products. And so we see a net new marketing channel, which is all of these AI LLL models. And we've seen the traffic directly land on our product details page, which we, often abbreviate to PDPs, way more than than past. So in the past, people would either land on a product listing space like rings or necklaces and then go on, use the on-site filters to find the product, whatever catches their attention. They go on the details page. But we are now seeing the share of, landing traffic on the PDPs, significantly increase, week over week, month over week. And I'm not gonna go into year over year because as we all know, this has completely changed as as of last year this time. Yeah. Artin and GK, that's so interesting. I think one, the idea that, you know, again, we we think about interacting with an AI experience as, again, it's it's an it's an AI. It's a it's a bot. Right? It's not a human. But I think our team talked about how because it is natural language query, it's inherently a more human way to interact with that product, that content, and that brand, which which I think is really interesting. Right? It's it's much more human than, again, clicking seven different links to then get to that PDP, right, at GK. And I think the other thing is, I mean, that that direct traffic to the PDP page is so interesting because I I know just how how much better that converts than someone who lands on the home page and who then has to find that product, all on their own. So it's it's it's really interesting. And I guess it it it makes a lot of sense, you know, that, this idea that you can use AI as almost this advanced search, this thing that gets you to where you want to go much more quickly. Now where those folks wanna go more quickly, I mean, that that seems like a lot of that is dependent on understanding your customers at at a deeper level. Right? It's much more about who they are, what context have they provided you, what are their intentions, what are their past behaviors. So I'm wondering if we can dig into that a little bit. How are you thinking about customer segmentation in this new world, and how do you shape that different experience to that different buyer? Because, again, that entry point is so much different than it was even a year ago. GK, I'd love to hear from you first. Yeah. 100%. And as you said, Joe, I mean, it's it's very much that earlier we all the brands, any d two c brand had an opportunity to introduce their brand, introduce their companies, as you said on the home page where you could inject content, but you were not still talking about product. Let consumers figure out their own adventure from there and, interact with the navigation, go on to a listing page, and then whatever products catches their interest. All of that, you can bundle it up into a ball of paper and throw out of the window because that's not happening anymore. What's happening there is so in terms of segmenting the customers, I'll I'll take a very there's several different ways the way we slice, and segment customers, but I'll take a very simple explanation, is prospect customers who are the customers that we defined have never bought from majority, and then existing customers are the customers who have bought cost who have bought from majority. I can go further down into level two and level three segmentation of, for example, prospect, someone who's visited the site but never bought. So there's a repeat visitor but not a purchaser yet, and then there's a first time visitor. And then for the existing customers, they're active customers, they're dormant customers, they're champions who, who have a much higher LTV, who have a much higher repeat purchase rate. Once we segment all of these customers, the way we are shaping experiences is now if that first time visitor is landing on the site, that first time visitor is not landing on the home page anymore. So we don't we don't have the benefit of introducing ourselves as a brand to that first time visitor on the site through the traditional channels anymore. Now we need to surface that content on directly on the PDP because they've had that discovery journey outside of our website in in their, AI agent chat. So how do we then segment those customers and surface the content and build that content is the ongoing conversation. And and we've started to do that a bunch of different ways. There's always, different experiments that are running on the side, but that's just an example that if, let's say, between the three of us, Artin is a first time visitor on the site and he lands on, a product details page. He'll see a whole piece of content very differently that nobody else sees, because now we need to introduce the brand. Now I need to tell Artin that New Jersey is a brand that have more than 65 stores across the globe. Build that trust, build that credibility with him. While on the other hand, Giovanni has already bought it from Juri. So he does not need all of that context, which now in Giovanni's case adds that as a distraction because all he wants to do is look at the product, fall in love with the product, look at the price point, understand the product details, and and then go through the purchase journey. And so it's a fine balance of how do I not add more distraction for my existing customers while still building the context and the credibility and the trust with, with my prospect customers who have never bought from us. And so that that's the whole conversation and iterations that that we keep going week in week out. Wow. That's so interesting. I have a lot of thoughts about that, but I I wanna hear from our team. How are you thinking about customer segmentation and different experiences for, for your own users? Yeah. I wanna actually start with, like, a, like, a, kind of, like, more, like, a high level claim is that it feels like right now, it's, you have to resist the temptation to add things. And, like, what I know is my business, like, it's so easy to, shape new feature. It's so easy to, start adding, like, sub, like, a subsegments within your main segments, like, because the granularity of the data is so much higher. And the insights that you can derive is also, more, you know, granular. And, but, yeah, like, to Jake's point, like, we do have, like, a behavioral kind of segmentation. Like, and in our case, it would be the, like, installation, activation, what are the completion rate? Like, how many, like, courses or podcasts user created or listened. So, but we also have, like, a rich profiling. And the rich profiling kind of helps me to understand those behavioral patterns. Like, what kind of profiling intended, like and then it created the archetypes. Like, what archetypes have a completion, like, higher completion rates? What archetypes have a lower completion rate? And because in our case, the prompting is so rich with data, creating this profiling dynamically and in a some way, even unsupervised when we allow the others to actually derive new, properties of the user object to kind of enrich it. Those things became, really possible, and it would, like and it's completely was impossible before. Now this profiling, layer is, very good not only for understanding how you can create more personalized experience, which is very important, right, for the conversion rate and JK, you kind of cover that all. But it's also then a little important for targeting because you can see the behavioral patterns or patterns of your users. You understand match certain behavioral patterns, and then it helps you to identify and prioritize what target users you wanna go after because they display those behavioral patterns on your platform that you actually want to achieve. And, that, like, a deep level of the intelligence and insights, that's what fascinates us the most. Behavioral, yes, but more importantly for us is a profiling. And our profiling goes, of course, like, a demographics, like, about the users. But, we also, enrich it, with, some, like, adjacent things. Like, for example, we want to learn, like, whether or not the user has a family, and that does a that the user listens in the car, our podcast, by himself or herself or with the kids while user drives kids to the school. And if user listens that with the kids while driving to the school, it, and the user like, some of the users prompting that, like, hey. I'm with two of my kids. They're these names. And we see that there is a car play signal, and we have a assumption that user might listen it with the kids, and we can prompt the user, are you with, like, alone or are you listening with your family? And if you press it with the family, within the content, we start mentioning the name of the kids and making, like, completely new precise experience that otherwise wouldn't be possible before. Of course, in our case, and we have the slogan. Right? Like, you are not a consumer. You are the reason content exist. Because, the LMS and, like, AI doesn't only transforms the, like, the discovery and browsing experience, but it also transforms the content itself. So in our case, the profiling clear, almost, like, something that's, you know, been unlocked that, like, a few years ago wasn't even possible. Yeah. Wow. So our our team, I mean, between, you know, rich profiling, these archetypes, being able to tell whether someone's in the car with their family or not, I mean, these are all things that I feel like give you a very, very, very good understanding of your user. You talked a little bit about targeting. You talked a little bit about experimentation. How are you using those insights and that rich profile information, to change the way that you build product that's on? So, like, so as of today, our platform, to be completely honest, is still as not, liquid as I would like it to be in terms of, like, truly creative experience, one to one for the each user. But this is something we, of course, actively discussing, like, how the platform can be different from the each for the each user. In terms of the targeting, as I said, like, you know, for example, if we know that the like, if we have a good understanding of the profiling and we have a good signals from the behaviorals, behavioral patterns within the platform. Right? We can indicate which users actually our ideal customer profile. And Mhmm. Even if I want that in your hands. Right? Like, once you know who you need to target, like, who is that, persona that, enjoys your product the most, it significantly simplifies your top level, right, like, the experimentations. Because you don't go just, like, super horizontal across different assumptions. You're more targeted. And every time when you have a more, like, when you have a more accurate or more precise targeting, it one way or another, it affects your cock, like, client acquisition cost. And once you can, like, lower client acquisition cost, then you have, more, room to experiment with your lifetime value. And, like, in terms of, like, the your pricing strategy, so you can start lowering prices and increasing the funnel. But, you know, like, but again, like, our profiling is the center, like but if I don't have a profiling, which I can identify my ideal customer profile, I will not be able to target as well as possible, like, today to bring those users on my platform. And also, another thing is, like, an an artificial intelligence is this concept actually, very, like, invent in the LMS and the AI in general. It's exploration exploitation. Right? Like, how much you exploit data that you know versus how much you explore new adjacent, possibilities. They are, of course, high risk, the low probability for the success, but they can increase your chances to arrive to this global optima to even something better than what you have today. And with the intelligence layer of LMS, actually deriving those adjacent exploration layers with your targeting, with your profiling. It's also much easier because it's not just a human sitting and building those assumptions. You actually can feed a lot of data, unsupervised data. And then, like, LLM start, giving those assumptions, which, allows the move much faster, experiment much faster, and quite often get much better results. GK, is is AI also accelerated the way that you're able to, experiment and also marry your understanding of your users to to impact how you experiment and what you experiment on? And today, you get 300%. Go ahead. And today, everyone says that what is the mod? Like, in a like, I'm not sure what I'm saying. Just mod. Right? Like, mod mod is the speed. Like Mhmm. And, like, how fast you move, how fast you iterate, how fast you because we all are at the end of the day, we're doing search. We're searching for the customer. We're searching for the good experience. We're searching for the value, to deliver, and, like, the speed is the big, like, positive outcome in that search. Yeah. GK, I think I think you're about to weigh in specifically on on maybe maybe speed and and execution around experimentation imagery. Yeah. Absolutely. And and to your question, Giovanni, I I wanna say 500%, or whatever number big number you like. Like, AI has completely changed on how we experiment, the speed at which we experiment, and also, more importantly, what we experiment and with who we experiment. So I touched upon a little bit on, the content generation for different kind of customers' segments, but then how that translates to, some of the features that we build on the site. So another way that we look at our customers is is also what is it that drives those customers on the site. Some customers go for design led products. Like, they get attracted to the design of the product, and they're not interested in in your staples jewelry, like a stacker ring or regular looking hoops earring. They're more looking for edgy design. Whereas another segment of customers go for they look for value. Another customers of, segment of customers look for product information. And you'd be surprised at how deep those customers want to go and learn about the products, different materials, what percentage of each metal and alloys and stuff like that. And so all of that, first of all, with, AI has the speed on experimentation is massive because we are able to generate prototypes what used to take at least a few days. First of all, if that was the number one thing prioritized first with the design, then we would build it into an experiment, launch it. Today, we hooked up our, design system with Cloud Design. The product manager is creating that prototype in less than an hour. And then they are able to take that experiment live within the same day. And in order to get stat sig, whether it it depends on how important the feature or how important the area of the website is, is you do wanna get stat sig or a positive, strong signal is good enough for you to roll it to a 100% of the traffic. And so all of these conversations has suddenly changed where we are no longer waiting for any every experiment to get get stat sig. And so that is on the speed of the experiment. And then going back to the different types of customers that I was talking about, like, one of the experiments that we're running on the site right now is for those product, knowledge hungry customers, there's a video pop up that pops on the PDP where a person is explaining the product in greater much greater detail than what you otherwise see on the on the PDP because that segment of customer is interested in that level of depth on the product knowledge. Whereas, to Artin's point that he mentioned earlier, for another customer, like, we have so much content, but the more important thing is how much we edit out. That level of depth and content would be completely irrelevant and rather a distraction for completely different segment of customers. So that's the way we are looking at all of these. Like, who is the customer that I want to be experimenting for? What is it that I want to be experimenting with? How am I experimenting? I I touched upon it briefly. And then, content pieces is something that we've carried away. So all facets of building experiences that includes the feature as well as the content and how we experiment, who we experiment for, has completely turned upside down for us. And so I'm I'm hearing a lot of really interesting things. I think one, the idea that you can go from from idea to prototype in an hour. Second, the idea that, you can have a richer understanding of your users and you can start to derive some of these signals. Right? So who's the the conscientious buyer that really wants to understand the trace of every mineral that or every metal that's that's in there that's in there, you know, the ring or their necklace or their bracelet. The idea of speed as a moat, which I've definitely been hearing that idea echoed quite a bit, and it's definitely a theme that we're hearing is the idea that, you know, there there is no moat, but there is staying two to three steps ahead of your competitors. I mean, with all of this change, right, we're shipping more product. We're all running more experiments. We're all collecting more data. But speed without signal is is just chaos. Right? And so, GK, I'm wondering how you actually measure what's working, and how does that change the way that you use data in your day to day role? Yeah. 100%. I think what has been traditionally defined as success criteria is important than ever. We can't anymore go forward and attach five different metrics and say all of these metrics should be positive when this experiment runs. We have to prioritize. We have to be very laser focused on what is that one single metric that we're trying to move here with which user segment, and for how long, and what does that guardrail metric looks like. So just taking that same example of a video pop up on on the website, we are only measuring that success for the segment of customers that we call, like, product knowledge. And we and the only success metric there is, just time spent on the PDP, whether they interacted with the video a lot. And another one is add to cart. And one of the guardrail metric is just bouncing off the PDP, and that is it. So we have one primary metric that we're trying to drive. We have a consequential metric that if that has a positive impact, sure. Amazing. And then we have one guardrail metric to also help us know when to kill the experiment that it's starting to have a negative impact. So that focus that I'm talking about is is becoming sharper and sharper with every experiment that we are shipping. And, also, because experimenting itself has become easier, we're also finding ourselves, to make sure that we're experimenting. So there's how in, I I wanna say, even two to three years ago, a lot of those features would be say, okay. We have significant experience in what used to be called, like, product sense and and gut checks. Product sense is still very, very important to know what you want to be testing about, but it's not But the gut check is is completely out of the window anymore because there's no reason why you should be gut checking anything when you have so much data at your fingertips. Mhmm. Arteen, how does the approach to measuring success, how has that changed for you? Yeah. Actually, like, the so there's a few things. So, like, I I think what you measure as a success, it changes based on the stage and phase of the product. Like, for example, when you rolling out the new product, right, you want to get the signals around, like, whether or not this is something people would love to, you know, have, like, or buy or use. And then, like, on the later stage, we want, like, figure out, like, in our case, it's whether or not it can be become a, habit. Right? And now traditional ecommerce, and by the way, like, I probably should put that in my introduction. Like, I've been in the commerce since 2011. I had my commerce store. I work at Walmart Labs, and I was at Adobe Commerce department. But, like, so, like, they're, like, it's, like, how you can do, like, a return customer, and you might measure, like, how many return purchases, especially depends on the, like, average order value. Right? Like, if your order value is low, like, you you actually, like, quite often waiting on the, like, return purchases versus the first purchase. And, I think it depends on the, like, product cycle. But, the way what changes is, I think, actually, branding and emotional connection, which is sounds a little bit like vague, but, they're actually more important than ever, especially when you are competing in a such a great market where there's, like, a so many, products, like the the the the marketplace is just so overcrowded these days. So, like, how you can stand out is double down on the quality. Like, communicate the quality and build a brand out of it and connect with the users emotionally. And anything that, like, can, and with the emotional connection, of course, the, like, return sessions or return purchase are good indicators. But also, we actually start focusing more on the qualitative feedback, as well integrating. So it appears. Of course, we're measuring the, like, you know, all type of the quantity of signals. But, qualitative feedback, and there is, like, this, like, famous, question for the product market is right. Like, how many of your, customers would feel bad if your shop will close tomorrow. Right? Like and, like, product market fit indication is, like, forty percent of the people who indicate that they will actually feel very bad if you close your business tomorrow. That indicates that you have that emotional connection. So how you can deal that so they don't feel that they interacting with, some, you know, cold, corporate, like, machine, versus they feel actually there's some, there's some sense of belonging. There's a some sense of, like, maybe community or some sense of, like, some feeling that makes them feel happy revisiting your store or revisiting your website, revisiting your app. I think those kind of things are important to nurture now. Yeah. Because because otherwise, if you're competing just in the future parity or pricing or if you just compete on some, you know, like, dry utilitarian value prop. Like, margins going down and, like, it's just too hard to compete in that, especially these days. Which is so interesting because, you know, we mentioned earlier how that that entry point to how someone actually interact with your brand, maybe that's no longer the home page. Maybe you have a little bit less control. Maybe it's through content generated on social media by a TikToker, not someone who's actually a part of, your company and your brand. So it's a it's an interesting challenge, and I'm really curious to see how, you folks will use, you know, data to figure that out. But I do wanna move on to what I think is a a really hot topic right now in ecommerce, and that is agentic commerce. I'm sure we've heard a lot about how in a few years, no one's gonna be shopping. I'd have sent my robot shopper out to do it for me. It'll be my personal assistant doing things on my behalf. I'll just, I'll just give it my credit card. Right? Do you buy it? And if you do, how are you thinking about just AgenTek Commerce in general? GK, let's start with you. Yeah. 100%. You know, that Agentic Commerce is definitely the buzzword of the year. So but but I actually ask a follow-up question, and that that's my personality. Like, when someone says and that's a product manager and me speaking. If you say it will it be delivered, I would say it depends. Right? Yeah. And so when someone says agentic commerce, my follow-up question is what exactly do we mean by agentic commerce? Agentic, is it like an on-site clippy type assistant or a chatbot where you see an assistant and you ask them questions and then they only answer your targeted questions and then you're on your own journey? Or is it more like the AI agents chatting journey that we were chatting about where people chat with their favorite LLL model about their needs, about I need a new ring under $300 and then go through the options, but then they have to go on the website and completely purchase themselves. Or as as you said, Giovanni, is it us giving our credit card but then asking our LLL model to go ahead and purchase this on our behalf? Or is it a level ahead where if your agent, through your activity, sees that Giovanni has bought a suit, he's bought a new pair of, formal shoes, this is a bracelet that will really go with this look and goes ahead and proactively buys that bracelet on your behalf. So that's my question. Like, what level of agent ecommerce are we talking about? And I think that answer will depend from industry to industry and from brand to brand. For Mejuri, specifically, I don't think we we'll at least for now, we we see it quite far away, whereas an agent would someone would feel comfortable with their agent going ahead and proactively buying something there because it's a pricey purchase. Not everybody feels comfortable, with AI shopping on their behalf. But we are certainly bullish on the fact that, once you've had that product discovery journey with your LLM model, you give them your payment details once you've made your decision, and they will go ahead and check out on your behalf. That is 100% going to happen tomorrow, if not day after tomorrow. So that's definitely the scenario or the version of agent ecommerce that we're bullish about and we're definitely building towards. Artin, you've been thinking about a building in commerce now for for a number of years. What's your perspective? Oh, I I I agree with j JK. You know, like, there's a certain topic experiences that, will not be disrupted. Right? Like, if I, like, if my, wife, she likes to go to Lululemon and put things on and, you know, kind of, like, have that experience. Right? I don't, necessarily buying the idea that that experience will be disrupted in the next two, three, maybe five years. But, I can totally see, agents becoming more proactive as a assistance on finding things and, you know, with your confirmation, perhaps completing the purchases. I also can see the world where the, like, the commodity purchases, like, toilet paper, like, a soaps can be somehow, you know, like, abstracted to the agents, because the, because the other the experience of, like, discoverability and, you know, like, and the shopping itself is not as, important for many people, right, when you are purchasing, like, say the shoulders, shampoo. Like, so, I can see that. I also can definitely see, like, on the digital realm, like, like, agent to agent purchases, data purchases, access, And, I totally can see people, like, setting up credit cards and purchase limits for their agents to proactively go conduct certain, you know, like, creative work or, digital researches, or, things from that nature. Would it be less on the, like like, shopping jewelry. Like, I don't think I would ever ask my agent to go and purchase, you know, like, necklace or watches for my wife. You know? Like, maybe, maybe the like, as the element of search but again, you know, like, maybe, like, you're going through, like, all local shops, see what they have prices, like, price comparisons, like, more on the utility realm of things, not, actually, the, like, selecting or, like, experiencing the shopping for me. Yeah. That's that's very interested Mhmm. Yeah. On that note, I do want to mention one more thing, Giovanni. Like, the product manager and me won't sleep well tonight if I don't mention the trade offs about going agentic. Right? So one one of the trade offs is all of us remember a time when we were all about blocking bot traffic on the website. And when we are analyzing the site traffic, how do we exclude the bot traffic out of that analysis so that we can get a true picture of what our site visitor looks like, the health, and and all the fun metrics that all of us have come to know so far? And now where do we stand with all of these agents crawling on the sites at any point in time and a lot of those agents being deployed by us as individual users on our behalf to to take actions on these sites? And that makes that, like, contradiction very real today. The other piece is in in an earlier part of our conversation, we touched upon profiling our customers, having a deeper deeper understanding of our customers. Now if the customers themselves, through their own actions and through their own browsers and cookies, are not taking actions on the website, where does the data for that profiling come from if it's only the agents that are taking actions on on our behalf? So it's it's kind of a, like, chicken and an egg story or or a spiral cycle story, where if we go bullish on one end, there are definitely trade offs on the other versus if we stay conservative, then there's there's always someone else, ready to disrupt if if you don't do it yourself. So those are the fine balances that we have to think through. And and not to age myself, but I also recognize with with the generation where if it's a flight booking, I'm not making that booking on my phone. For that, I have to whip out my MacBook because it's a big purchase. I need to validate all of all of the details. And and so it also depends on again, to my back to my point about what industry, what, brand, what price level are we talking about, and the comfort level of each generation interacting with that brand, in their own way. The new generation that is AI native, which means that they've completed their high school and their graduation with AI somewhere in the background, will act and respond completely differently to to a generation where we are adopting AI in in a part of our career, but all of our education was pre AI. And and so that will also change how a lot of different brands respond to, all of these, progress points. Excellent. Excellent. Excellent points. We'll we'll end it on on, I think, a a short and and sweet, question. Right? I think we've we've we've been talking about all the things that are going to change from the way that we experiment, from the way that we consume data and profile, our our customers, from the way that we set up brand experiences that deliver the value of the brand and gender loyalty, even if that person doesn't land on that perfect landing page you created for them. But, Arlene, I'm curious. Like, what what doesn't change? What are the fundamentals that still hold in this new world? Brand relationship with your customer. You know, like, this actually, we have this, like, discussion internally, like and actually, Jeff Bezos, I think, very popularized this question. Right? Like, focus not on what gonna change, but focus on what not going to change. And, and then, like, Amazon case. Right? They're, like, focusing on the pricing, fast delivery, like, as the core of their and, like, a wide assortment. Right? Like, you can purchase whatever you want. People always will love to have selection. Although there is a good argument against that too. The paradox of choice pool covers the whole not that the, like, choice paralysis and so on and so forth. But, I guess in our quality in our case is, we double it down on the quality of our content. Like, we want to, and we're, publishing benchmarks, like, constantly, like, how we perform other against others, proving that how it's gonna do the highest quality, how we deliver. Like, want to deliver fast. Like, the experience has to be fast and especially, I guess with the actually, with LLMs, people started having more, tolerance. So because you can have, like, a deep search and you have to wait, like, five minutes. So people actually have tolerance for the thinking process, but still speed, like, quality, speed, customer service. Like, again, he like, I have nothing against the automating customer service. But, if you do that, and especially if you're a big brand and I was at Walmart Labs. Right? Like, I was in a couple of then and I was an intelligent business growth team. So, like, if you automate it, it's fine, but make sure it works well. Right? And make sure it truly delivers value and customers don't feel that they just been abandoned. Like, every time when I go on Google, like, services, workplaces, and I wanna, like, connect to their customer support, I feel that no one really cares about me there. So, I feel like customer support is actually a good chance to win those ambassadors, those true, lovers of your brand, because they already have a emotion and they can switch from negative to positive. So yeah. Like, a quality speed, customer support, that will be my bet. I don't think it's gonna change anytime soon. GK, what do you think? Yeah. Arden summed it up pretty well. I see it as, like, brand. He perfectly summed it up on the brand, but the other thing is economics. People will always care about what they're getting for the money that they're spending. Just because an AI agent is spending on their behalf, it doesn't it's not like people will stop start caring less about what they're getting in return. So and and that's just pure economics. So a combination of brand economics and the customer service that they're getting will not change, whether it's an AI agent or a human taking an action. Awesome. Well, thanks again, Arthine and GK, for your insights. We're super lucky to have amazing customers like you who are, I think, not just at the forefront, but so willing to share your customers like you who are, I think, not just at the forefront, but so willing to share your thoughts and your insights, around how other companies, I think, navigate these, these fast changing, landscapes. We did our best to answer questions in the chat, but, folks, if you have questions, feel free to reach out directly to myself, connect with our team and GK on LinkedIn, and we will also send a recording out after the webinar. And so thank you so much. Thanks everyone for watching, and we'll see you next time.