How TUSHY Achieved 85% MORE Sales with Gorgias

How TUSHY Achieved 85% MORE Sales with Gorgias Speaker 1: Early on, we just knew education and answering any wild question instilled trust. And when people trust you, they will share their hard-earned money with you. A year ago, I was like, okay, I need a sales team that can really sell baby sell. But it's awkward for humans to be like talking about, you know, your hemorrhoids and then be like, And would you like to buy a 36-pack of toilet paper? Like, that is not the way humans communicate, right? And then this tool, it very eloquently leads people through that journey. They know it's AI, so I think they're asking questions that they might feel judged by a human, and so we see like 85% more sales come with the AI sales agent than with our human agents, which do have a high conversion rate, but it's a different path, you know, and maybe it's much more direct than our appuros might be. Speaker 2: Welcome back to another episode of Chew on This. Today we bring to you a special series brought to you by Gorgias, where we're going to be breaking down how customer experience can actually drive your brand growth. We have the CTO and co-founder, Alex, as well as the Senior Director of Customer Service, Ren, from TUSHY, a brand that you probably have in your household. So before we go down and break down some of the really cool tactics we're going to be sharing, I'd love for each of you to give an intro. We'll start with ladies first, Ren. Maybe for the few viewers who don't know about you, give them a little bit of a breakdown of what got you to TUSHY and what you're working on today. Speaker 1: Yeah, so I joined TUSHY in the wild ride that was 2020. I was building an online travel marketplace, very aptly launched during November 2019, our IP vacation, and I was Reflecting on how I was going to pay my rent, my husband jokingly said to me, I was living in Mexico City at the time, and he said, you know, they're having a toilet paper crisis in the U.S. There's a shortage there. I bet you could get a job for TUSHY. And we laughed about it, but I went on the website and I was like, oh, my God, they're totally sold out like early March. And I was like, oh, man. They must be having like the time of their lives with their customer experience right now. And then two weeks later when it was like, oh gosh, this is going to be a real thing. My husband spotted a job posting on AngelList for customer service rep. I sent an email in that day. I think the subject line was when the shit hits the fan. And I was like, so I was like, you know, I'm wildly overqualified for this job, but I can hit the ground running. And if things are as bad as they look on the website, which the website looked great, there was just nothing to buy. I can help you and I was on the phone with our now CEO in half an hour and I was in the queue the next day and it was a total shit show. But what I learned was I'd never seen brand love like that before. Trying to build a community around surfers. And I'd had a coffee shop before that. I'd worked for a local grocery startup that all had really strong communities. Here was a product that you couldn't get. We were telling people on the website, this is going to ship to you in June, in March, and people were forking it over. And I was like, there's something here. And when I joined, it was a very small team. And I've been there a little over five years and have had the time of my life. And but yeah, had some pretty big problems to solve. Speaker 2: That's incredible. Thanks for sharing that journey. Otz? Speaker 3: Before I start talking about me, I just want to say I'm a customer of TUSHY. I got two, highly recommend getting one. It's great. For me, we started my co-founder Roman back in 2015. At the beginning, when we started, we had the Chrome extension, pretty much like that was the beginning of the journey. And the Chrome extension was just like right faster on the internet, pretty much with templates. But that allowed us to discover that, well, there's like a lot of support agents out there that want to automate their customer service. We spoke more with them. We went to a couple of like accelerators. One of them was here in New York, and we raised the seed round. And that allowed us to kind of start building the product that a lot of our customers use today, and some of them love to. And yeah, so since then, I mean, we've been growing, but maybe I'll share one story where I remember Roman And maybe it was like, maybe eight years ago or so, he was just like, we got to build a Shopify integration. So I was like, why? He's like, well, I'm talking to all of these merchants and they use these other products, which I'm not going to name, and they don't like them. It's not really adapted for what they do. And I was like, okay, let's look into it. Oh, by the way, you have to build also the billing thing. And I was like, We already have a bidding system. Developers just say no by default, that's just the nature of it all. But he was like, okay, just look at my leads and it's like eight out of 10 were Shopify merchants. And I was like, okay, yeah, data wins, data wins. And so we build it and I'm very thankful that we did because now we're here. And we're able to support about 15,000 merchants or so, including Tusha, which I'm... Speaker 2: And Avi. Speaker 3: Yes, of course, yes. I apologize. Yeah, and I would say it's like There's a lot of learnings along the way, happy to share as much as I can. Speaker 2: I'd love to jump in on a couple of fronts. So I think one of the biggest pieces that, and I'm gonna jump right into it, is like you look at where AI stands today, right, and the cultivation of it and how it started as like something that was being built in this tech hemisphere and now it's like Moved into sales, customer service, everything. I'd love to share our example of it. Ren, I'd love for you to maybe talk about how you guys have integrated, but maybe before we start to give our examples, Alex, you can kind of just tell us how you guys have adapted AI. No, without any bias, I'd say like from a tech integration to like usefulness, like AI for you guys compared to any other tech enablement tool we've been using, probably one of the fastest adoptions. Like you guys Double down, you made sure we really got behind it to where now it's like, I don't remember a world without it. And it's only been, you know, a few months. So it's really cool. I'd love to understand what's been your vision to drive that home. Speaker 3: Yeah, 100%. It's interesting because I believe yesterday there was a leaked memo or so from Tobi from Shopify, which is pretty much saying you got to adopt AI. I mean, that has been the theme for at least two years, right? But what was interesting there was you don't get more headcount in your department unless you explain why I cannot do it, which I thought is quite interesting. Yeah. I would say it's like, I think pretty much. Maybe not as explicit, but a lot of companies adopt that approach right now. When we look at the applications that are actually working today in the industry, it's like, oh, you can do image processing a little bit better. There's certainly a lot of memes right now with ChatGPT and all of that. But really, business use cases, I would say customer service is a very good fit for it because it's highly repetitive. The knowledge is usually available online already, but sometimes there's a lot of it so that the customer has a hard time processing everything. So it's definitely a good fit. The other fit, I would say, is on the coding side as well. It certainly accelerates a little bit the software development. It's not perfect. It's maybe intern level, if you want to have an analogy, but it's quite capable intern. And I would say maybe this kind of summarizes a little bit like the, you still have to supervise it in code. However, on the agent side, which is, let's say, like more autonomous, which is what TUSHY, for example, is doing, and you guys as well, the autonomy is like end-to-end. There is like post-factum supervision. So like people are reviewing things rather than, so there's like high enough trust to actually respond and communicate directly with the customers. That increases like response time that like also, maybe it's like even if the AI doesn't fully resolve your ticket, let's just say, it still collects a lot of information beforehand, such that when a human agent looks at the ticket, they already have a lot of information. So it's like, yes, maybe it's not fully autonomous, but in all 100% of cases, but already does a lot of a lot of good things. And then there is like, when you look at the agent specifically, they typically have different skills. So you have a customer service skill, and okay, like the agents can respond, where's my order type of questions. But what we are developing right now is the sales skill. So if people, for example, ask, like, hey, I want to buy bidet, but I have no clue what I need, right? So it's not a complicated product, but there are certain questions that you need to answer before you buy, right? And so historically, what TUSHY was doing is like they had this very elaborate Kind of step-by-step guide where you click a button and it clicks a second one and third one and to some extent it's still there, the wizard is still there. But it's just like the conversational aspect is such like a natural thing today. Everyone uses ChatGPT, Perplexity, like Claude and so forth. So it's just a natural kind of progression to use that and it can recommend products as well. It can understand products, like we just rolled out a much more improved... I'm curious what you think actually. It's just out of the cooking stove, I guess, if you will. And it's like performing so much better, like we essentially doubled the number of purchases after we improved that and it just kind of gets better and better every day as well because also It doesn't work without data, right? So it's like you gotta feed it data and it needs to be done in a thoughtful way. I can also say a couple of negative things that also don't work today, I would say. The thing is, when you're building a product, when you're inside the business, and I'm sure that's the case for you, there's a million things that don't work. And from outside, it can appear as like, well, it seems like everything is working well. But when you're in the inside, a lot of things don't. And I think one thing that I would say the next iteration of these agents is, Exposing more the inner workings and how they actually work and what is the reasoning steps that it takes. So you can understand if something goes wrong, at least you can understand what happened and improve things and stuff like that. So I would say that probably is, let's say, the next stage. There's other things as well, which is also around AI, which is not necessarily related to agents specifically, which I think are key. And I think we're going to see more of that, but maybe I'll share that like a bit later. Speaker 2: Absolutely. On that same note, I'd love to know from your end, especially from like the tactical space of how you guys have taken, especially this new tech, how you've implemented it and what kind of impact it's created. Speaker 1: Yeah, I'll go back before the Gorgias tool when we first, like, I'm a total nerd, I'm a lifelong learner, very curious, and when I got my hands on ChatGPT, I was like, oh my god, this can, like, work magic, and we were, a couple years ago, we were in a space where people were using it, like, secretively, right? Like, oh, my boss doesn't know, it's responding to all my emails, and I was like, I've got this team of really hard-working people, but they're not always at their most creative at 3 a.m., right? Their time. And so most of my team is abroad in the Philippines and so I was telling them like lean into this tool for the things that you want to do, particularly creative writing, right? And so we had this customer write in in Shakespearean verse and Cheryl, our Puru, we call our agents Poop Gurus, she was like, how do I respond to this customer? I was like, Add it in the ChatGPT and see what it comes up with. And she was like asking for return beyond the window and manifested into like six to eight emails back and forth in Shakespearean verse and like for sure the most delightful experience this woman has ever had. And Cheryl's like, can you believe that she keeps responding to me? And I was like, this is the power. Of AI, it's not replacing people. And that was a really beautiful lesson for the whole team of like, AI isn't coming for your job, it's going to help you do your job better. And I want to uplift my team in that way. And so yeah, after that, like I had a colleague, he was like, I just can't get this work done because I don't have enough people and I've got to write these job descriptions. I was like, ChatGPT it bro. And he was like, ah, no. And it was like early, like, no, you know, we don't want to do this. And He wrote three job descriptions in 20 minutes and then he was like hiring these people and he had like it's this kind of thing of like humans are still gonna be required but using it to agilize our work and with the instep of AI with Gorgias radically changed our workflow of like we are a very heavy Q4 company typically like around the holidays, Black Friday, you know people decide and that's changing. But it used to be I would have to hire holiday help and I would start in July and by October they would be like ready to go and then you know I would keep the good ones and it would be clear like this is just holiday help and I always felt terrible you know like oh sorry your family's counting on this but like at January like it's probably gonna be over and this year was the first time I didn't have to do that and I fact I lost two teammates before that and had the most successful holiday ever because of AI and I was like Dave, my right-hand man, senior manager, we were like, okay, let's brace ourselves. This could be, like, really crappy and we'll just get through it. And then it was like, wait, our first response time is 17 seconds? Like, in, like, peak lead-up to, like, Claude? Like, okay. Speaker 2: That's powerful. Speaker 1: That's awesome. Speaker 2: That's really cool. Speaker 1: And it alleviated a lot. Like, my team, you know, they're all, they were afraid and I was like, We are not replacing you. We're leveling you up and like you guys are here because you have heart. The bot will do everything perfectly. You guys are here to compassionately care for our customers and you know teach them how to toddler train their toilet to use a bidet. Speaker 3: I would say it's like typically organizations the way you see it's like they typically have levels. They have level one, support level two and they do a bunch of escalations. I would say it's like what this tool does it's like I automate some part of like a lot of level one, some part of level two, you know, not everything. But already it's like it can help you save a ton of time, of course. Speaker 1: Yeah, I tell my team, like, if your goal is to be an email responder, this is not your place anymore. Like, I don't, like, I just, I have a very high bar for who makes the greatest of Puru and I want to be constantly helping them improve their career path and their personal development and I think it's like, make sure your life has meaning and answering Wismos all day long, like, it's not giving your life meaning. Speaker 2: No, I love that because I think what you, what you pretty much summed up pretty beautifully, which is, You want to reserve the time for your agents to actually handle the big needle movers, right, for the customers. I'll give you an example at Obvi where I would say like a good amount of the tickets that come in are like the normal BS, right? Speaker 3: Where's my order? Unknown Speaker: It says it's delivered but it's not here, right? A lot of that can be handled with just AI, right? Just right off the bat like, hey, we have your information, here's your tracking, check X, Y, and Z and you should be taken care of, right? So the response time on that, perfect. It kind of reserves the more, I would say, in-depth questions that we're getting, especially around supplements and health and beauty. It's like, well, I'm dealing with X, Y, and Z, and everybody's X, Y, and Z is way different from everybody else's. How do I use this product? You know, I'm working maybe reverse hours. I have, you know, a job at night and I'm sleeping in the day, like things like that. At least having our agents have the time to at least thoughtfully now address these problems, right? When the rest of the, I don't want, it's not nonsense but it's the more monotonous requests come in, right? And so to your point, you know, our customer service agents know our products really well. They know the ins and outs, the benefits, the ingredients, everything. But every scenario is going to come in and it's going to be way different. And that's why I think even like leveraging ChatGPT to say, you know what, I don't want to overstep and say something that I'm not supposed to say because I'm not really a healthcare worker or somebody like that. But can I submit this scenario to ChatGPT? Give me a response to that will not again not overstep but give a proper response back to the consumer in a way that's personalized specific to them. And not only is that going to obviously kind of solve the issue. Helps with the retention factor. Not only can you start maybe upselling a few other things there, and it saves on churn. So I think that's the biggest factor where we've learned. And I'm curious how you guys are leveraging this to kind of focus on those core things, which is turning customer service into more of a profit center versus, you know, where's my order and kind of responding to that. So what are some of the core things that you guys are focusing on to use this to become a profit center? Speaker 1: Well, this AI sales agent is doing a beautiful job. I think the thing for TUSHY, you know, when I started working there in 2020, by and large, people didn't even know what the word was, right? Or how to say it. So like, our core job is education and pushing culture, right? I think, you know, we've sold over 2 million bidets, but that's still, you know, 360 million people to go, right? Like, there's a lot of people that are wholly unaware. And so, We've found, and the data reflects this, that it's education. It's not sell, baby, sell, right, buy this, because people are like, I'm 70, I've lived my whole life without this, I do not need this, you know, it was like... Speaker 3: Oh, you do. Speaker 1: Yeah. And like we equate it to like a cell phone, once you get it, like you can live without a cell phone, but your life is much more enjoyable with one, it's easier, and like bidets, like people don't usually go back, but It's a long road to get there. People think it's gross or it's weird or, you know, puritanical ideas about what it is providing you. Like, it's just a water diversion device. It's like cleanliness, right? Nobody questions the shower. And so we found that answering all the TMI questions or like, you know, setup questions or just all the weird cultural stuff that people are like, The weirdest question that every member of my team, they're like, Ren, why does everybody ask how to dry themselves afterwards? Like, because you guys grew up in the Philippines, where washing your butt is par for the course, you never had to think about it. But these are people like with internet access, credit cards, and the joke is like, what kind of panic attacks do they have when they get out of the shower? But and so a lot of it was just through education, like what do people need to know? Like, install is maybe the biggest bulk of our queries. But then it's also like, how do you get them to actually install it? They bought it, but it's like still this thing of like, they don't know that they need it. And so sharing about, you know, health concerns, and again, touching on your point, like, Not necessarily from my mouth, but like we found that our consumers, our customers have found relief for these kinds of things. And so early on, we just knew education and answering any wild question instilled trust. And when people trust you, they will share their hard-earned money with you. And so it's a slow game, but now we've seen over time, you know, we were one of the hot products in 2020 and COVID, right? Everybody got a TUSHY, they got a Peloton, and I don't know what else. But bidets are one of the only products that lasted during that time when people decided they're going out of their houses, you know, everyone's, you know, sold their Peloton on Facebook Marketplace, but bidets stood the test of time because it was just truly this cultural shift. And so the question before was like, why do I need a bidet? And now the questions are, Which bidet should I get? Is it the attachment? Is it an electric bidet? Is it an analog seat? Like, which one do I want? And so, yeah, it's really exciting to see that journey. And this is where making that shift, like a year ago, I was like, okay, I need a sales team that can really sell baby sell. But it's awkward for humans to be like talking about, you know, your hemorrhoids or your kids like anal fissures and then be like, And would you like to buy a 36-pack of toilet paper? Like, that is not the way humans communicate, right? And so to ask people to do that was really hard. And I was like, how am I going to do this? Like, they'll just be a sales team. And then this tool came in and it's like, it very eloquently leads people through that journey. But also people know for our bot, they know it's AI. So I think they're asking questions that they might feel judged by a human or they know it's a robot. So they're like, and it's, Unabashedly selling them on these things and it's just a different journey. And so we see like, yeah, 85% more sales come with the AI sales agent than with our human agents, which do have a high conversion rate because of all the knowledge that they're sharing, but it's a different path, you know, and maybe it's much more direct than our purers might be. Speaker 3: Absolutely. I would say it's like also functions all the time, right? Speaker 1: Yeah, it's always working. Speaker 3: That's another thing. For me, what was surprising when we released the initial version, and thanks for being an alpha customer, but what was surprising to me is just like, this thing sells better than I could. Because I thought it's probably not going to be that good. I was very skeptical at the beginning. I mean, by nature, I'm an engineer, so I'm skeptical by default, but I was quite skeptical at first. And the other surprising thing, I guess not very surprising, but it turns out that selling and doing customer service is very close to each other. And what I mean by this is like actually you need the kind of similar knowledge because people ask questions about to buy products and they ask questions about products, how to install. And so it's like the information that is needed to sell, by the way, it's also needed to do support. And I would even argue that selling is slightly easier. No offense to salespeople. One of the capabilities that we have is you can, for example, let's say cancel an order or that type of stuff automatically. And that needs a connection to maybe Shopify or whatever ecommerce platform you use. And that's kind of like a little bit more complex to set up because you need to like do the authentication, authorize it, like all of that type of stuff. It's not that hard, but still like you have to do extra setup. On sales, it's mostly just having conversation, sharing information, giving opinions and kind of asking a lot of questions like leading to a sale, right? So addressing concerns, maybe offering a discount. It's a lot I would say it has a high ROI in a way and but also it's like it's a lot easier to set up the other thing that's interesting is If you're paying, I don't know how many dollars to get all the visitors on the website and a lot of people buy stuff on the weekend or during the night or whatever in a different time zone, I don't know. And your team is maybe not there to respond that immediate second, you know, because people are super busy. They don't stay on a website forever to wait until like kind of people have the time to respond. And so I would say it's like, hey, if you want to increase your conversion, You should have, maybe not Gorgias, try something else, but you should have something to respond to your customers because I'm willing to bet you're probably losing some conversions. And I think it's, you know, maybe we're not going to talk politics here, but I think today, especially, it's very important to be efficient with the cash that you have on hand. And yeah, that would be my message for a lot of people who are listening who have an ecommerce business. And I'm sure it's top of mind and you have to adopt AI, it's here. You have to do it in a thoughtful manner, not just, oh yeah, just enable here and it's fine. You have to invest your energy in it and you have to believe that that's kind of the path forward for your company. And not only on support or sales, but pretty much across the board, right? We have to be more efficient. I hope things are gonna get better very quickly and we're gonna, you know, maybe forget about this, but I think it's like it's just better to be cautious and kind of lean into it. And I think there's always risk with adopting a new technology, but not doing it is also very risky as well. I can talk about Gorgias and say it's great, but it's just like, guys, you gotta do it, even if it's not with us, you gotta do it, because it's really important. Speaker 2: Yeah, I'd love to maybe chat on kind of what we can call it, maybe myths or things that maybe are a stopping point for people, right? And I'm sure maybe you have examples that can tailor to it, or Alex, maybe some circumstances. But I feel like the one thing that people are always saying are like, I just don't know if I need this. I feel like I can just do it myself. Some people just are like, I have trust issues of knowing what this AI is going to say. What if they say the wrong thing? Mind you. Speaker 3: What if people say the wrong thing? Speaker 2: And so, you know, I know we have a good section of viewers or listeners that do maybe fall into that bucket or just like, I don't know if I need this. You know, and so like maybe give us your thoughts. Speaker 3: I think that's fair. I think you should be skeptical. And I think people tried versions of it. Like, I mean, when we started in 2015, Around that time, it was all about like chatbots on Messenger. Remember that? It was all about like, okay, this is like the conversation or whatever. That didn't turn out to be like a thing. I think and a lot of people since then, like veterans in the industry is like, yeah, it was not working then. So like for example, I mean, we've seen what ChatGPT 3.5 did and it was like not so good. And then GPT 4 came out and everyone's like, oh my God, like this is amazing, right? And even like that version is probably not amazing by today's standards. It's probably crap compared to... So I would say it's like the past experience is not necessarily the, you know, oh yeah, I tried it a year ago. Things move so quickly today. I mean, I spoke about coding a little bit, like how quickly that kind of transforms. It used to be like maybe, I don't know, an eighth grader a year ago, and now it's more like an intern. You still have to supervise so that stuff doesn't... Engineers are still on call. When stuff breaks, the AI is not on call. That's what I keep telling my engineers. But it's too much. You can't ignore it anymore. So you should continuously try And there's going to be failures. But the good thing is like we really thought about it since the beginning to address these concerns. Like one of the things that TUSHY has set up is like kind of Topics where the handover is immediate, which are very sensitive. So I'm not going to say which they are, but you can set that up. If you feel concerned that, you know, I don't want the AI to answer on this type of questions, you can certainly set it up. And I think, yeah, it's a common thing right now. Speaker 2: Yeah, I think we have an example like that where it's like if somebody's asking about an ingredient and you're answering it, but then they're like, oh, well, I'm getting headaches or I'm feeling this way or something like that, then you can. Speaker 3: I believe it's about control. So there's like a lot of, let's say guardrails that you can set up and that's true for pretty much like all the tools out there. So that's number one. But the second one, which is really, really important is because there's going to be situation, things are going to fail. Like the AI is going to hallucinate. Well, guess what? People also hallucinate. So that's going to happen, right? But you can ask a human, like, hey, why did you write this? Like, oh, I thought that, you know, like we still have a discount on this particular campaign, for example, or whatever. So it's like a failure mode. It's pretty difficult today to kind of dig into all of the complexity that led to an answer today because these AI agents today are like pretty elaborate already. Like they are not just one question and one response. The way ours works is like there's many, many different large language model calls like that kind of all Conspire together to give an answer to the customer. There's like a validation step at the end so that there's no like, I'm going to give you a discount for $1. I'm going to give you all my stock, that type of stuff. We put a lot of effort to make sure that the AI doesn't respond in a weird way. But that also, because there's a lot of guardrails, sometimes it's harder to understand what actually happened. And I think that's one thing that where people are still feeling like, oh, I'm not sure if I want to adopt is because if something goes wrong, I should be able to explain to my CEO what went wrong. But if you're too careful, then you're missing out on a lot of benefit too. So it's like, I think my take is that these agents Throughout the coming years, they're going to be a lot easier to, quote unquote, debug. You'll be able to understand what they're doing and they'll be able to explain the failure modes better, which will lead to a lot more trust. I believe it's like, at the end of the day, it's this, right? It's like, if you don't trust the system, It's like people's like, hey, if I have a decision tree, if I have a tree, it's like I trust it, but it doesn't do much, but at least I know how it works. And I think ignoring this and telling people that you should still use AI because whatever, and ignoring the fact that they have legitimate concerns, it's not the right approach. A lot of work still to do but you should definitely try it already today and because I think and you should probably even if your let's say business is not affected immediately by like some recent changes. You should still kind of get ahead of it because you don't want to get into a point where it's too late, too little, you know, so it's like what it's actually like burning, you know, so. Speaker 1: That's what I would advise everyone. It's like we're on the bleeding edge of this thing. I think in our world it's easy to feel like everyone does it, everyone's talking about it, but like ask your mom what she knows about it and you'll get a different answer. And so I think it's Making sure you're aware of what's out there, deciding if it's right for your business, but at least educating yourself. Because yeah, otherwise you fall behind. People are afraid of the internet, right? And we're all employed by the internet now. And like, my Nana, you know, she does not shop online. And I'm like, it's kind of limiting for you. You know, she's trying to buy all this stuff to remodel her home, and she needs to go in store. And I think I understand and respect that. But I also think, yeah, Be educated and then make the call. Speaker 2: I think one of the reasons people are probably skeptical is cost associated, right? I think when people really look at it and they're like, oh, well, I have to spend X amount to kind of justify this or I need to make X amount from it to justify it. But I do think there's another way to justify it and it comes from the assistance that it gives in terms of Reducing the amount of people that actually cancel or return products, right? Speaker 3: Or converting customers that are not yet converted. Speaker 2: Exactly, right? Unknown Speaker: I think there's three buckets, right? So the people that haven't fully bought yet but are like on that edge, then you have the people that have bought and then you have the people that have bought but want to return, right? And so I think probably one of the lowest hanging fruit is the people that Want to return, right? They were obviously sold on the product initially, but they weren't fully, fully sold, right? And so I'm curious on that standpoint where are there any retention tricks that you guys are doing at TUSHY that maybe, you know, with the help of AI, people can implement that maybe you couldn't do before, but some of the strategies to kind of reduce churn. Speaker 1: Yeah, I think our problem was the converse. It was more like getting them to convert because they were so skeptical of it. So I feel like AI is really helping us on that front because our return rate is incredibly low. Thankfully, it's typically under 2%. And that is where our PURUs still focus. Many return reasons will go to manual review. So if it doesn't fit my toilet or I'm having install trouble, all of those go to manual review because The number one reason is doesn't fit my toilet. We're like, we know if it's like 95% of toilets, so that's okay that you're not a plumber, like let us guide you through it or send you this part that you need. And so that's where we're flipping the manpower and pushing them to try it because it's like if you can return it in 30 days, no harm, no foul, like at least try it. And so it's because Now we're in this new era of we've converted people to bidet users over the last five years, thankfully, most by and large to COVID, right? People couldn't get toilet paper, they discovered bidets, there was mass adoption. But now we're in this phase where we're up leveling them to the next product and so I think that's what's really cool. They might come to us from a landing page or a meta ad or something and they learn about only one product and that's what we've learned a lot is people don't know that we have a full product catalog, that we sell toilet paper, that we have toilet scrubbing brushes, that we have bathroom scents. They just came to a lander, they bought the classic which is you know our entry-level model and it's one and done. Most people in New York they have one bathroom, that's one of our biggest markets. But when AI gets in the mix, it's like, well, you could get the classic, but what are you actually looking for? If you could, you know, what's your budget? What are you looking to solve? What's your bathroom look like? Oh, you should absolutely get the Aura. It's got a heated seat and it'll like sing you a lullaby practically. Speaker 3: It has infinite time, infinite attention for you. You're like the only person in the universe for it, right? Speaker 1: And so, and visibility to the other products. And so our Subscription model for toilet paper is robust, but it's still a small portion of our business. So thinking about when we talk about churn and retention, it's really thinking about how can we lead them to repeat purchases. So the average American household has two and a half bathrooms outside of New York. But also like, it's a really great gift. So maybe you only have one bathroom, but like, and those are the things that the AI can really tell it like, oh, you got a housewarming coming up, like this is like gonna make you like a king amongst your friends. You will never feel bad about going to the bathroom. You never have to leave TUSHY behind. So I think it's thinking about that from a different angle, because this is, you know, You might only buy one in your lifetime. So it's equating it to like, you know, a fancy coffee machine or, you know, maybe a rug for your living room. You know, these are not things that you're buying all the time, but we want every person recommending TUSHY to all their friends. Like I had a phenomenal experience with them and like, I had X, Y and Z also happen and they had my back every step of the way. And that's what we're trying to instill with this tool. Speaker 2: I'd love to spend some more time, Alex, especially from your vantage point. You've seen now across 15,000 plus merchants, probably seen everything under the sun from a customer service point of view. Maybe you can share some really cool creative examples of how some, an example of either where it's where AI has come in and relieved an extreme amount of work or creative efficiency or just really some creative ways where you've seen CX being turned into that Sales agent kind of overnight. I know for us, just real quick, similar to you, during Q4 we were trying to ramp up to see if we're going to need some more help and till this date, that's when we had looked into adding the AI piece, till this date we have one full-time person and then one part-time person and during Q4 Expect to go to three, but with the age, I mean, forget the cost savings. I mean, just the efficiency of not having to manage new hires and new people to help and this and that has been huge. So curious, are there other anecdotes and examples? Because I think sometimes people only relate to seeing, oh, well, that could be me. Speaker 3: Yeah, I mean, Ren touched on like the holiday season. It's certainly just like kind of sourcing people, doing interviews, hiring them, onboarding them. That's a lot. I mean, it's just like, okay, there's the salary, right? Like per hour, like the seasonal part. And we've seen also like kind of BPO, like agencies doing like adopting more AI because they also say it's like, hey, like there's some elasticity. We are going to have, we're always going to have some people that are doing maybe level two work and they are trained to use this AI. In fact, I would say it's like agencies are actually becoming sort of the experts of AI agents and setting them up. Because they understand that's where the direction is going and they want to make sure that they're still in business. And guess what? They're becoming one of the best people to set it up. And certainly, the product feedback that I'm hearing is also coming from a lot of There's definitely efficiency, I think that's not a question. I have one anecdote maybe with one of my customers that I see, another one, where they didn't use AI before and they just responded with a macro to a question. Someone bought a gift for his brother that passed away, unfortunately. And the response was a macro, which was like, just, okay, they wanted to return the product, it was quite expensive. And so, I mean, the personal response is like, hey, like, okay, I get it, like, I'm going to follow this. But I feel like that's just a missed opportunity. But I also don't blame the human agent, because the reality is, is like, they have to respond to hundreds of tickets, if you don't have an AI setup. They're going to have to respond to hundreds of tickets. There is no time for empathy. They have to do it because they have that quota that they have to keep. It's not like they don't want to or they don't understand. Sure, people want to, but it's so limited if you don't do it. I use that example quite often. It's like the notion that the AI is somehow not empathetic. And it's not capable of adopting, let's say, a way of speaking. I think that is a big surprise that we all had collectively, right? We used to think that AI is this sort of robot thing that is doing robotic things and it doesn't have this kind of creativity to it and so forth. But it's actually quite capable of adopting a lot of human qualities today. And I think things are only going to get better in this direction, which is great. But I would say empathy is hard to scale. That's why when companies become very big, you get that like, oh, yeah, I'm going to be one hour on a phone call. It's just horrible. So, I would say, you know, you should do it if you're emphatic with your customers as well, because it certainly can, like, handle, like, you know, hard situations. Just like saying, like, I'm sorry to hear about your loss. Speaker 1: Yeah. Speaker 3: It's huge. Like, it's just... So, anyway, that's, I mean, it's maybe an extreme example, but I would say it's quite representative of what, like, these systems can do today already. Speaker 1: Yeah, I concur like it kind of blew my mind in the same way of Wow, this is like a perfect response and we've even used it in calibration reviews like when we look at tickets as a team like This is what a perfect ticket looks like because, yeah, the bot doesn't get burnt out or tired or doesn't have a kid that's sick with a fever and doesn't have to meet the productivity quotas and I think, yeah, a human can always see like, oh yeah, I missed the mark there, but the bot is like checking a list and it's, yeah, pretty phenomenal and We are very transparent that it is a bot, but we've seen that in reviews too of people being like, wow, I am so impressed by this, because we had a lot of trepidation of like what it would feel like for customers, and we're kind of in this middle ground where some people still say like, You shouldn't have given me an immediate response because it's clearly AI and we're like, it also says TUSHY concierge bot, like that also tells you, but I think we've also found like people like the immediacy and the accuracy and it can be very empathetic and compassion, particularly around these kinds of things. My dog ate the package or I was in a car accident and it will always say something that we haven't told it to say that we've just said be empathetic. Speaker 3: Absolutely. Yes. I mean I can definitely confirm that and also on the sales side as well and on other things too. It's like the other thing that I would say which is quite interesting and like I have this sort of personal vendetta against CSAT. Speaker 2: Because I think I remember that. Speaker 3: So anyways, like, there's 15% of people actually respond to CSAT responses. However, if you're automating 60% of your customer service, it does mean that like a huge gap in the ability for you to see if your AI is doing a good job or not. I would say it's like for people too, but like people just don't click on those responses most of the time, right? And usually it's very polarized. It's like either one or it's five, right? But then you look in the five stars and then you see that the agent is giving discount codes, so it's kind of appeasing the customer, maybe a bit too much. So I think the other aspect of AI is that they can be extremely good tools at analyzing the quality of the interactions that the people have. Something which I'm really excited about, which is coming next, is the ability to understand and classify what people are actually talking about. And that's kind of like our voice of customer initiative that we have. So we are building a way for the merchant also to understand this, like, hey, this queue generates this amount of tickets and customers are talking about this thing there and you're generating some of the revenue with it as well. And there's basically all these dimensions that are really valuable for the merchants to understand. So basically it's like, It's not just about responding to customer, it's also about understanding what the customer is talking about and making business decisions. So that's kind of another iteration of this. And it's not only, yeah, only text and that's it. They are very good at other things too. Speaker 2: Yeah, someone said I think this is like every day what you look at AI is like the worst it's gonna be. Like it's only gonna get better and then you look at it from that lens and it's like, wow, like that's the real power of starting early. Speaker 3: 100%. Speaker 2: Because, you know, it only can adapt and understand better. I think to your point, it's like if you asked me five, six years ago, it's like, well, what do you think AI can do for you and your business, right? I probably would not have an answer because I would have zero idea what the capabilities could be. So putting you on the spot, Ren, is over the next couple of years, where do you want AI to go and where do you hope it can help you with your current workflow and how can it make things better? Speaker 1: Yeah, I don't think we're far off from this. I want it to be able to analyze images really effectively. A lot of the questions that we get are, will this fit my toilet, for example. Yeah, and video, and we always need a photo or video. I mean, like, I'm an expert through and through, but I've been on the phone with the customer like, yeah, the doohickey doesn't attach to the thing. Oh, yeah, I'll totally send you the thingamajiggy. And then it's like, two, you know, seven to 10 days later, they're stoked on the call, but then they get a part that isn't the right thing. So like, For years, we've had to always ask for photos or videos. This is like a rigorous process to get them to do that. And there's better tools out there to do that. But to be able to analyze effectively, like, yes, this will fit your toilet and these will be the models or assess installation scenarios like, oh, I can see clearly this is what's going on. And so like, I don't think we're far off from that. Speaker 3: We talked about it. We talked about it just today. I'm not kidding. That was one of the topics that we had on the product meeting. Because, I mean, GPT-4 already supports images and all of that type of stuff and is quite good at processing images. So, it's coming. It's just, again, like a natural step forward. I mean, voice as well. That's another channel. As you said, Ron, it's the worst that it's going to be. Yeah, so I mean I know because I look sometimes at some of the tickets and so I know that like images are... Speaker 1: I was like, if we tell them every time like position the toilet and this thing, you know, like the eyeglass try-on thing, you know, I was like maybe if it's like this it can confidently say, but yeah, I think that would change things. Speaker 3: But also image instructions too, right? Because like you get the image but then it's like, okay, so what do you do with the image? You can describe it with text, but you can also give examples with images too. So that's like another... Speaker 2: It's crazy. Speaker 3: Yeah, yeah. Ultimately, what I've seen also is the agent is pretty much trained similarly to if you hire someone who is external. And so if you invest in kind of documenting the things that you're doing in the business, it can be used for, even if you don't believe in AI. It's like, I'm never going to use AI, whatever. It's still valuable for anyone, like a person to use it. Fair enough. Speaker 2: I mean, I honestly don't know where the time went because this was an incredible episode. I wish it was a little longer because there's other questions I have. But I would love, you know, from the brand side and from like on the operation side, two things, you know, what would you want the audience and viewers to kind of take back from this episode and implement in their business today? Speaker 1: I would say lean in. Early on I said, you know, you have to break a few eggs to make an omelette. Thankfully there wasn't that many broken eggs. But I think it will allow you to get to know your customers better, but also do things that you might not have been able to do before. I think it's really important for any CX team to have a really strong product marketing feedback loops and this is what it's allowed us to do. We have weekly meetings. Because bandwidth has been freed up and we have better data, right? Data. And it allows me to be ruthless with my data. Like when I can track everything, I can, instead of just being the loudest voice in the room, I can say like, oh, this is the ROI on this or this is why we need to be fixing this thing or adding this product because of what I can show now. Through really, yeah, concise tracking, but also looking at things As our business grows, I'm like, you know, I'm on a ticket level now, but like if I got a hundred thousand tickets in a month, what would that look like? And it's not possible to look at every ticket. And I think that's where AI comes in. If you want to grow your brand, you have to be learning it now because you won't be able to scale later. Like a hundred agents, like you can do that, but you'll just be losing touch with your audience. Speaker 3: On my side, what I would do if I'm a, let's say a business owner of ecommerce brand, like what I've seen I don't think this just applies to ecommerce, I think in general. I think, first of all, you need to tell your employees that it's okay to use AI, it's completely allowed, and it's okay to make mistakes. And when they happen, it's not okay not to learn from them, of course. You need to make sure that It's completely fine. You should use it and you should kind of go after it. Then what I've seen work is whoever is on the middle management or something like that, they need to show maybe specific examples for their team of what works. And identify whoever is the, let's say, the person who is the most, let's say, advanced in the organization. It can be anyone. It can be a support agent. It can be someone in operations, like whatever. And have that person kind of be in front and say, it's like, hey, I've used it like this and kind of show. So encourage people, but also show how it's used because a lot of people don't even know how it's used or they know like various things. Another thing that I've seen works is that you essentially kind of create someone who's even, like give someone a position who's even like a chief AI officer, if you will, like kind of, I don't know, like whatever AIs are in the company. And their mandate is essentially to go after- Smart, very smart. And it's just like, hey, I will go to every single department and try and see how can we apply. And I'm going to spend a lot of time researching the tools out there, right? And so that's what I would do. Basically, it's like, can you identify someone who is just a natural sort of person that already does it? And that way they can spend a lot of time making things more efficient and advertising what's available there. And I think it's just going to become a habit at some point. People are just going to, why would I, wouldn't I use this thing? It's like, it's been, it's been something I've been using for the past five years. So of course, like it's a normal thing. So anyway, that, that would be the very tactical. What can you do right now?