#705 - AI Search Optimization for Amazon Sellers

#705 - AI Search Optimization for Amazon Sellers Speaker 2: Considering AI traffic is up 527%, OpenAI is running for orders within ChatGPT for native checkouts, and ChatGPT is already driving more than 20% of Walmart's referral traffic, the AI ecosystem is definitely one you will want to master, which is why today we bring on a special guest, Joanna Lambadjieva, to talk to you about the core data sources that are behind these large learning models and how you can make sure that your products are staying ahead. They're showing up where the customers are searching for them. Unknown Speaker: How cool is that? Pretty cool, I think. Hello, everybody, and welcome to another episode of the Serious Sellers Podcast by Helium 10. I'm your host, Bradley Sutton, and this is a show that's completely BS-free, unscripted and unrehearsed, organic conversation about serious strategies for serious sellers of any level in the e-commerce world. Speaker 2: AI has obviously become a very sexy topic as of lately and it's changing how products are discovered. It's changing how consumers are discovering their products and actually buying into those products, right? And so what we're going to really be discussing is how the search engines come about discovering those products and how you might be able to do a little something if you have a business of your own. Without further ado, let me bring up Joanna. How are you? Speaker 1: I am good. What an intro. Thank you so much. Speaker 2: Tell me about you. Tell me your background. I know you, but the viewers might not. So let's get into that. Speaker 1: Yeah, sure. So I'm Jo. I love that Shivali is using my proper Bulgarian name. But for people who have met me, I'm just Jo. I'm originally from Bulgaria. But I've had like quite an international journey to where I am. I spent about 10 years in the UK and about seven years in Germany. So I'm kind of, I would say I'm the citizen And, you know, I've been in digital marketing e-commerce for about 14 years now. And out of those, like, at least 10 has been solidly in e-commerce. And for the last two and a half years, I have been kind of obsessing over AI and how it will make like our lives and also our professional lives completely different. And so yeah, that's kind of me in a nutshell. Speaker 2: Amazing. Well, I know you have a spectacular presentation lined up for us. Unknown Speaker: If you want to watch the Serious Sellers Podcast in video form, make sure to go to YouTube and check out our brand new YouTube channel. Just type in the search bar Helium 10 Serious Sellers Podcast and you'll see our S logo come up. Make sure to hit that subscribe button. Go to the channel and binge watch any of our episodes that we have done lately. We'll see you there. Speaker 1: One of the topics that I talk a lot about, I write a lot about is basically how AI is going to redefine search, essentially from a consumer point of view and also from marketing and e-commerce point of view. So today I'm going to talk a lot about that. I am very respectful of people's time and generally I like to earn my trust before I start talking. So I want to tell you why maybe you should spend the next 45 minutes With me listening to me and listening to what I say. So when I was saying that I'm like obsessively looking at AI for e-commerce, I'm not lying. So for the last two years, I have been essentially every single week reading more than 20 to 30 different sources in order to compile a newsletter, which I write three times a week. And I'm super, super passionate about this topic. And I think there is a lot happening. And I feel like, you know, I spend a lot of time to compile this. I've written more than 300 issues on the topic. And I feel like that has really resonated with audiences. And as a result, the newsletter has grown quite significantly. But also, as I said, I'm really passionate about education. I like to share what I learn and to like, because I really believe that AI is going to have a massive impact on all of our lives. And I feel that everyone should learn how to use it for their business and just generally to prepare for the future. So I basically on my LinkedIn, I posted more than 50 SOPs today. Which I feel like have resonated also because I've grown quite significantly there in terms of followers and LinkedIn has thought that, hey, this content is useful. So somehow it's gone viral. So it's really cool. But I am showing you this maybe as a humble brag. I think you can definitely see it's a humble brag. But more so just to show you, I guess to like prove you that you don't have to assume that I know stuff, but I want you to see that I really am passionate about this. I'm a real nerd on the topic. The way I built this entire strategy is from a position of reading a lot, researching a lot, and experimenting a lot. From that position, the one thing I'd say about 2025 is that in just generally e-commerce and in Amazon, we are not in the year of hacks anymore. I've come from a long period of just focusing on Amazon strategy, And, you know, there was a period where, like, people were sharing all kinds of wonderful hacks and they were working. But I feel like now we're moving into a different era of Amazon sales. We know that right now the buyer's journey is much more complex. It's not just about humor anymore. It's about the customer. And so part of the piece of, like, growing on Amazon is not just about being on Amazon, but to think off Amazon. Not to think about the future. And for me, like from everything I've been seeing for the last two and a half years, since I've been obsessing about AI, is that the future really holds AI search everywhere. And this is this is like a massive trend. And I think it's going to impact us much like sooner than we expect. So this is actual research that came from Ahrefs and SEM Rush about two, no, this was actually Search Engine Land, about maybe a month ago. The first piece was the fact that AI traffic is really, really growing. Actually, like the stat came out, basically, AI traffic is up 527% SEO is being rewritten. Now, of course, in the like massive scale of things, when we think about Google, this is still very, very small percentage. But although in terms of absolute size, it's not necessarily comparable to like the classic Google traffic. What we're seeing is now with research coming out that actually all of this traffic is very small, is very high intent. So like this research from Ahrefs basically found out that only half a percent, less than a percent of their visitors, of the visitors coming from AI search engines, drove more than 12% of signups. So this is something that I'm seeing across various industries. Basically, people who come through AI search engines like ChatGPT, Perplexity, Claude, Google AI Answers, they have very, very high intent because just generally the amount of research that goes into their click is much higher. This is even more recent. This is actually news from this week. So like the first piece of news and I wrote about this yesterday is the fact that OpenAI is actually reading like a whole orders sort of native checkout within the environment of ChatGPT. And just as a, it's kind of like a, like a numerical perspective. Right now, ChatGPT has about, I think, 700 million weekly We are talking about ChatGPT, which came out two and a half years ago, if so. The scale with which this is growing and the impact that it's going to have by having a native checkout platform within the environment of ChatGPT is going to be hugely, hugely disruptive for how we run our e-commerce efforts, how we think about marketing, how we think about consumers. Another stat that actually I'm going to write probably in the newsletter tomorrow is this stat, which is ChatGPT is now 20% of Walmart's referral traffic. This is really huge. So, you know, like the evidence kind of piles up and piles up, but the point is like AI search is not a problem that we're going to have maybe You know, three years in the future, this is something that we need to actively start preparing for starting thinking about, because it's not going to just impact like, you know, let's say, just need to see, but it's going to impact those brands that are selling on Amazon, because the general like the entire sort of search funnel or like search journey is changing. To one where you are like chatting with an AI chatbot, finding all this information, the AI chatbot goes and finds like different sources and then recommends your product. So very different than what we are used to till now. Alright, now the million dollar question. So how do we rank for these AI search engines? Well, as Elbow says, it's a bit of a black box. Let's face it, we are so much at the start of this, like this new era of like technology. And, you know, you can see it like a lot of people are putting out content. We're recommending different things. And to be fair, we are all figuring out like, you know, I don't think there is like 100% right or 100% wrong answer. I think as we go along, all of these AI search engines are also going to change the way they recommend products. But what I want to show you today is a little bit of a process, a method or recommendations that are based on the data that we currently have. What do we know about ranking? The first thing that we know is that ChatGPT loves Wikipedia. So the first thing actually that is worth knowing is that different AI search engines have different preferences in terms of the sources that they basically quote in their results. But we know definitely that ChatGPT overwhelmingly likes to quote or to like source citations from Wikipedia. So Wikipedia is very important. But what is really important also to notice is that the second place goes to Reddit and Reddit And Reddit is a very, very, very important source. And I'll show you why, because on the next slide, if we're looking at Google and Perplexity, they really overwhelmingly love Reddit as a source. So this kind of shows us that, you know, both Wikipedia and Reddit have to be part of our, let's say, optimization strategy in order to get these AI search engines to start recommending. Of course, like, you know, if we, like, dwell into the chart, we can see other sources. We can see YouTube. We can see Yelp, LinkedIn. Like, you know, the list doesn't stop with Reddit and Wikipedia. But if we have to make our life a little bit more manageable, at least in the short term, we're going to focus on those two. The other thing that we've learned from research about AI search engines is that AI search engines love reputable sources. So if we're talking about something like Reddit or Wikipedia, even Wikipedia, like Yelp, these are all kind of user-generated peer-to-peer like sources. And because of the freshness of the information, because of the fact that it's peer-to-peer, AI search engines love them, but they also love the forms of this world, the tech crunches of this world, the sites that really have proven to have a high, essentially high publishing reputation. And again, this is another research from AHRF and it just shows basically what AI overviews refer in terms of like, looking at like and recommending sources. And one of the really interesting things or the highlights of this research is that AI visibility isn't just about like, let's say, optimizing your website, but it is about how widely your brand shows up across the web. So it's not just about, let's say, being quoted on like one or two other websites. It's about the spread, like the brand spread of mentions across reputable sites across the web. So a lot of talking, a lot of theory. Now, what do we do about it? So I want to show you today kind of We like three ways or actually not ways the three kind of approaches to tackle each of the different platforms. Again, I just want to sort of We end this conversation by saying this is not about hacks. There is no magic pill. There is no just from one day to the next solution where all of a sudden everything is hunky-dory and you're going to start getting recommended just by pulling one switch. This is more about creating a process, methodological process to starting to build that presence across different places. So let's go. Let's first tackle Reddit. So what are we going to do about it today? So the first thing is we need to find the relevant threads in our niche that basically are going to be ripe targets for us to like appear in. And then we are going to try and as much as possible participate in these threads and mention our brand slash product. And then we're going to repeat and rinse, rinse and repeat on a regular basis. I mean, simple, but not so simple. But let's break it down. So the first thing, like anyone who has seen me speak before, one of my favorite tools when it comes to Reddit and just Reddit intelligence is called GovSearch. GovSearch is a brilliant tool because it in essence aggregates different Reddit threads This podcast gives you a lot of information about patterns in terms of what people are talking about, specific topics, anything that is kind of like a rising topic in terms of comments or upvotes or relevance. And so it's really, really good. But today I actually want to just take a slightly different approach because I find that SEMrush gives you a little bit more detailed data in terms of traffic. And I think that in particular when you are starting off and you have Very limited amount of resources or time. You want to make sure that you focus your time and effort on building your presence on super, super relevant and super high volume as well. The first thing you're going to do is you're going to go to SEMrush, and then you're going to type in just reddit.com. You're going to search for Reddit. And then within that, you also have the opportunity to use the advanced filters. So within that, you will basically include your main keyword. So in this case, I'm looking for any keyword that mentions joint and any keyword that mentions dog. So basically, anything that is related to dog joints. So obviously, I am in some sort of And then the cool thing is that basically what Stemrush is going to give you is for each of those keywords is going to give you the URL of all the different threads that discuss this. And they're also going to give you the different volumes and search. So the search volume is what will help you kind of see through what is probably quite an extensive list and to actually focus your efforts. So then the great thing is you can export this into a CSV. And now you have your CSV file with all of your keywords and all of the threads. Again, here, a little bit of like the duping work just to make sure that you have a very clean list. Now, what do you do from then? Well, the reality is that what you want is to essentially participate in these threads or like actually answer the questions that are posed. But please don't do like this brand has done, which is, you know, there is a Today we have a thread about joint supplements and somebody asked, oh, can you give me some sort of supplement recommendation for my dog? And, you know, there's a there's a brand called the Big Damn Dog Co, which says, definitely check us out. Our product was designed specifically for giant breeds, smiling face. Two months later, there is another one, like another comment right underneath saying, check us out. So this ain't gonna cut it, neither for the customer that might be reading this, nor for the AI search engines. What you really want to do is sound like a human, not a brand. Because again, it's really off-putting for customers to be sold to. Like, you know, Reddit is a community and this is really important differentiation from maybe like social media or any other place. Like Reddit is really a community. Every thread has its moderators, there are rules, there are strict rules for every, like, presence of Reddit about like who can participate, what is allowed and not allowed. And so you need to just In essence, understand that Reddit in itself is a universe and you have to play by its rules. And so what you want to do is essentially sound like a human who is helpful. And I mean, ideally, if you want to like attract customers, that is exactly what you should do. So, you know, you can see here that real users supposedly have written like quite a lot of sort of anecdotal examples of like, you know, what they've experienced, why it's helped their dog, why they are recommending this particular supplement. And it's all very, very human written. Like, you know, I believe that these are real people. And that converts, that is appealing, that is convincing because, you know, you want to get a real peer to peer recommendation. So what do you need to do really is to make your comments and make your participation really as authentic and helpful as possible. So how can you do that? There are strategies on what you can do. The first thing you want to do is indeed match the tone and the culture. And this is really important. So like, you know, you need to kind of feel the vibe. Oh my god, the vibe. But it's true. So like, when you respond, you don't want to sound like super professional, you don't want to sound like super formal. And if vice versa, you don't want to sound super jokey. You want to understand how the people and like how your ICP in essence talks. And a little pro tip for that is you can literally just You know, take all of the comments in that thread, put it into ChatGPT, and ask ChatGPT to analyze the tone of voice of this thread, and then based on that, write helpful answers. So again, you don't have to reinvent the wheel, but you need to be aware of how you sound and how does that fit into the thread. The second thing is, I mean, it's a no-brainer, but provide value. Sometimes, you know, I see a lot of people that just kind of go into a pure self-speech, and that just doesn't sell. What you really want to do is provide something that is super relevant to that customer, whether it's something about explaining a special ingredient and why that helps or, let's say, one of your customer's anecdotes, but provide something that actually pushes the conversation forward and actually helps that customer. And so, you know, once you then mention your product as a buy away option, that is far more, that's far less likely to get you like banned by a moderator, or like, you know, just, it's gonna be far less obvious than you just being like, our product is great, buy it, because nobody really gives a shit. Third is do not include links. Links are going to get you banned and just don't do it. You want basically customers to search for your brand. The fourth thing I'd say is that you need to be consistent. So like it's not a set it and forget it kind of thing. You need to ongoingly target these trends, like engage in the conversation, participate in the conversation, but also don't just, you know, post every day. Like, you know, space your presence like two to three weeks. So it looks actually like authentic and legitimate. So you just want to be mindful about the time and again, be authentic. And fourth is like I say is like, you know, use a seed and amplify strategy with a micro narrative. So like, what does that mean? It means that you're going to create kind of like a micro narrative for so like you're going to like say a short story like just what I what I showed you before with the Dane exam where a customer was talking about their Dane and like how he's been really struggling and la da da da da. You want to build like an actual picture around this narrative. And so once you have that, that kind of hooks the customer to actually want to find out more and what happened. So like, I'm just going to give you an example. This is the one I wrote. So I've been testing this is like for, like, for basically another year, I've been testing every term on and off for nine months, took it to heights to the gym and dropped it in it on concrete twice. Kept cold for 24 hours, no dents that affect sealing and the cap still doesn't leak. I wrote a tiny list of what actually matters when choosing one. Materials, lid seal, design, warranty, drop or reply if you want the checklist, plus a mini photo set. So you are kind of like, you know, asking for I think you're basically giving a hook, you're giving some sort of leeching. But this is kind of a way to get people to engage. And again, you don't this is this is imagine that you are not a brand. This is like you need to go into the role of an actual customer. And again, provide value. So yeah, summary, simple, not simple, like, you know, use tools like SEMrush or Gummy Search to really find out where you should be spending like that time and valuable effort. And then use, you know, use like ChatGPT to like help you inform some of the answers, but maintain authenticity. Again, like this is not a short term strategy. This is kind of like an ongoing process that you have to build within your organization. But I guarantee you over time, it's going to build not just within the kind of like the I work in the ecosystem of AI search engines. If you get a lot of mentions in Reddit threads, obviously people really read Reddit to find advice. Therefore, that is going to increase your brand awareness, your searches, and then drive actual sales on Amazon. Wikipedia. A system in itself. And to be fair, like the process does not make a sexy slide. So instead, I decided to like give you a nice SOP that you can follow step by step, which is actually going to be much more useful way to spend our time together. Alright, so let's talk about brand spread. Because brand spread is kind of one of those things where like everybody's like, you know, across all of these sites, what do you mean, Joe? Like, I don't have like, endless amounts of budgets. And also, I'm not like, you know, I can't get into both like How can I do that? All right, well, I will show you several options. The first one is around long-term DIY. So there are sites or there are platforms that essentially help you get connected to publishers. So there's three options that I put here. One is called How, the other one is Help a Reporter Out. And then there's two more, which are Quoted and Featured. So Quoted and Featured are more like pay to play. So you can, you know, for a certain amount of money, like you can get essentially featured in publications like Forbes, etc. But I first want to show you how you can use Arrow or help a reporter out to do it all for free. Again, it's a process, but it's like it's something that you can do kind of as a like a weekly task, if you wish. So what is Farrow? Farrow is basically a newsletter that you subscribe to, and they publish the newsletter twice a day. So they do a morning edition and an afternoon edition. Within the newsletter, well it's not a newsletter, it's an email, what they do is they basically provide all of the Today we're going to be talking about requests for sources, links from different publications and others. So, you know, and they split them by section. So, like, you have general, you have health and pharma, you have lifestyle and entertainment. And essentially, it's like kind of a great way for journalists to, like, go and ask for sources like, you know, can I get an expert in this? Can I get an expert in that? Can I get this and this and this? And for brands and experts to essentially match that. And so the great thing is that it like skimming through this will take you exactly three seconds of your day. But you can also in a way automate this based on like using Zapier or anything like based on specific keywords and AI. And this can be pretty automated. But the gist is, You go through the newsletter, you find a relevant thread, and then you basically answer to that journalist. So here is how kind of a query looks like. I'm looking to do a cat kitten guide on my blog. So I'm looking for items for either kittens or cats. And then they basically give you the specific Now, again, unless you have tons of people, I would say this is an ideal task for automation through ChatGPT or Claude or another AI platform. Here, I basically provided a little prompt, which, again, you don't have to use this prompt. It's just a prompt. You can make it much more diverse. But in the end, it's an easy way for you to build or write something that will probably match what they need. So I like to break sometimes prompts, especially when it comes to anything like... Let's say more information heavy into two stages. The first stage, I like to ask the AI to ask me enough questions in order to get the right information, in order then afterwards in stage two to write what I needed to write. So in this prompt, I basically said, stage one, clarifying questions. Before writing the response, ask targeted questions to ensure the content aligns perfectly with the request and the user's expertise slash products. And then ask, you know, your role and expertise and there is, you know, explanation. Product Service Details, Credibility Improved, Tone Preference, Call to Action, and Index. So this will, like once you answer these questions, ChatGPT or Claude will have a pretty good idea about how you exactly want to position your product. And then stage two, you know, I basically I instruct the AI to then go and write a quality compelling pitch that directly addresses the journalist's request, includes one strong hook, highlight what makes the product expertise unique, use short impactful paragraphs, contains one or two relevant links, and so forth. You can probably build something more tailored to your own brand and product, but I just want to show you how you can approach it so that you don't have to hassle with the actual writing. Alright, so again, pretty straightforward. But again, it's it's not a it's not a hack. It's a process. It's a process that like, let's say you or your VA does or like an automation does every week. And one thing I'd say is like, I love like, good automation and AI within like to optimize things. But if you're dealing with journalists, Ideally, you want to also be there, especially in the beginning until you start building a relationship. Yeah, to build a relationship because, you know, if you build a relationship with a journalist, especially if it's a really good publication, well, guess what? You're golden. So again, don't just delegate everything to AI. AI is great, but it's a tool. It cannot replace humans yet, and relationships definitely not. All right, quick pay-to-play. So this is the magic where you cough up some cash and you basically get launched in some publications. And no, it's not digital PR. Though, funnily enough, I thought PR is dead and now it's having a massive renaissance due to AI. Well done, guys. But no, this is a slightly different approach. So what I want to show you is essentially, firstly, how to find relevant sources which LLMs are being trained on. Second step is to actually validate this. And then three is to pay to write about you. Pretty straightforward. So the first thing I like to do is I actually try to, I want to formulate a hypothesis about where should I be present. Again, within every niche, within every specific product type, there will be different sources that ChatGPT or Perplexity or Claude will take in consideration to basically find information and then give you recommendations. And so it's very difficult to just say, yeah, these are the 10 publications. But the way I would do it is I would build myself a smart hypothesis about, OK, so how can I probably figure out which of these publications? What I found is that there is two quotes by Toro. And it has like a neat little function, which is audience research, where basically you can analyze a specific search term for a specific audience in a geolocation. And it will basically give you which are the most likely websites that consumers will read for that keyword. So here I basically have done one research like that for best ILAF serum in the US. And what it's given me is, in essence, kind of like an infinity score, traffic, domain authority, and linking website. Now I have things like Shake.com, Health, Marie Claire, Seventeen, Momsnet. So now I have a nice little list. Okay, so now, well, now, I will actually use ChatGPT to validate this. Of course, this is not science. This is just a hypothesis, but in the lack of data and in lack of like actually having a way to figure that out, I figured that it's a pretty good estimate to just go and ask ChatGPT on a scale of zero, not at all likely, 200, almost certain. How likely are the following websites to be used in the training data of LLMs like Gemini, ChatGPT, and Perplexity? And basically here, I've just dropped all of the websites I got from SparkTalk. So what then ChatGPT does is it gives back a fairly handy list of a score for each of those websites. And I can immediately see that shape.com has 75% likelihood, health.com has 80%, Marie Claire has 80, Seventeen has 75, Mumsnet 85, and so forth. So what I'm going to do here is I'm basically going to take anything that is above 70 as a pretty likely candidate. All right. Now what? Now you have a list of a nice website. You can basically do digital PR, do that. Or you can, you know, go and try and get yourself into these publications just by paying for them. And one service that I found that does that, which is super neat, is called PriceX. So basically what they provide in a nutshell is they basically get like press articles in these kind of publications on a CPC basis. So how does it work? So the first thing is like, you know, press exposes and pitches your product to publishers. So it's like actually pretty old school. But then if the publication is happy with this, the publication will write about the product. And then like anything that's like any sort of anything that you pay is based on a cost per click basis. So if you're actually not really getting any traffic or any clicks, you don't pay for it. And so it's kind of like low risk. And in a sense, it's a kind of a double whammy because, I don't know, it's double whammy, like a positive or a negative. Let's see. A positive double whammy because on one side, you're driving qualified traffic on a cost per click basis, let's say, to your website. On another side, you are also featured in these premium publications. And so here, for example, is one, for example, from Prestex, where they've gotten this keratin treatment in Marie Claire. And the cool thing is that they have a very, very extensive publisher list. So a lot of the publications that we saw on that hypothesis actually are here. And so I think for me, it's not just again about AI mentions, it's about the general picture of you also bringing that brand authority and associating your brand with a very premium publisher, which in itself It gives you lots of fun. So again, simple process. Go in and figure out your hypothesis with Spark Toro and use ChatGPT or Claude to validate them and then get in touch with Prestex and see what is available from publishers. I've created like a one-pager cheat sheet just so that I can simplify this entire process so that it's not overwhelming. And I've also kind of given a sort of an overview of how much everything costs. It's really nice to show people like loads of tools and then it actually ends up being super expensive. So here's actually what I want to show is that actually for most of these things, you don't really need to pay or you can do a free demo. So like SEMrush, you have a free seven-day trial. Power is free, SparkTor has a free trial. I think if you haven't paid for ChatGPT, please, that is your homework. That's the best $20 spent today, this month, this year, in your life, maybe not in your life. But yeah, go and subscribe to ChatGPT's 2025. You need to. And then press X. I think it varies by publisher. But now let's just talk about what you can do just on your website. We looked at a lot of like, you know, external players, but you can already just do something on your website. So this is one task for you next week. I don't know you, now you know me, but here you go, I'll give you a homework. So your task, your homework for next week is go and find your like basically highest traffic pages on your website or your like, you know, best listed pages. And then add a TLDR or too long to read. No, I got it. I didn't get it. But you know what I mean. And or a QA section to your product pages. So why is this your homework? Because basically AI crawlers love TLDRs and Q&As. It's basically how they try to be to basically grow all the information online, but also keep cost efficient. So the first thing is that TLDRs plus clear headings, let large language models pull like a very concise answer of what they are like, let's say what somebody has typed into ChatGPT without having to read absolutely everything that's on the page. So if you know the easier you make it for an AI crawler to like find the information they're searching for, well, guess what, the more likely that AI crawler is going to keep coming back to your website. Um, the second thing is that, you know, structured and parsable, like, love this, the lingo here, but like, just to make it like, super easy for you to understand is like, when you have a QA block, like, you know, what is this product good for? Or like, you know, is this product suitable for oily hair or oily face or whatever? That kind of structure maps directly into the model citation pipeline. So like, it's basically, again, about the ease of these crawlers to find information. Thirdly, it's the way buyers essentially converse with chatbots. So like, you guess where I'm going with this, but guess what, like somebody's like typing, what are the best $20 like, you know, hiking boots for very slippery, blah, blah, blah. And then when you have an answer for that, you know, this probably will very much match with like how customers are actually like searching or conversing with large models. And then like kind of maybe not for your product pages, but definitely if you have any blog posts or anything else on your website that is not product pages, make sure you do include dates, authors, citation methodology, basically anything that gives the large language model more data to citing because this again shows, I guess, trustworthiness. So How to do this? Well, you guessed it. If you guessed it, very good. If you didn't guess it, now you see. It's kind of very similar in terms of how you start thinking about optimizing for Rufus. And so how do you optimize for Rufus? Well, it's much more complicated. Then what I'm going to show you now, this is like one little part of the process, but you can use that part of the process to both do your website and your listing. So the first step is obviously use Helium 10 to get your customer reviews or We're going to talk about reviews and then use AI to analyze these reviews and then see what happens. Let's see what happens. So the first thing I'm going to do is I'm going to step this time, talk about ChatGPT. I know that everyone loves it, but I want to talk about Claude because Claude is really, really great at understanding text semantics and it's really good at understanding meaning. And so when I'm analyzing reviews, I much rather use Claude because it's just slightly more intelligent than ChatGPT. So step one is obviously get the Helium 10 review extension, pull your reviews. And if you don't have enough reviews, you can go and pull some of your closest competitors' reviews as well. The idea is that you want to have anywhere between 300 and 1,000 reviews in order to have pretty extensive amount of data. Then what you're going to do is you're going to just put those customer reviews like positive negatives into Claude. He's going to analyze them and then use this prompt. So this is a very neat prompt because it basically extracts far more than just kind of like a generic number of like Today we're going to talk about insights. You basically extract insights in a numerically representative way in which it basically counts the number of occurrence of, let's say, benefits or US fees or aspirations or negatives. So you can understand where the customer, like which are the most important or the most frequently mentioned benefits that customers mention in their reviews and therefore then use them later on to build on sort of like your optimization strategy and whatnot. So yeah, this is the prompt. But then what basically is going to come out of this prompt is like this prompt is going to give you this kind of table. It's going to be a four-column table with, let's say, the type of USP or negative, a summary of what people talk about, the frequency of mentions, and then actual customer quote. And then you're going to do this for each like benefits, negatives, or customer aspirations. Then basically this is how you build a knowledge library about your customers. So you're like, hold on, what are we talking about? Like, we're talking about AI search engines. Why am I analyzing reviews? Well, because understanding what customer buy this product for, what are their problems and like what they kind of like trying to resolve by using your product is pretty good indication of what are the most likely questions and they're going to ask Google's and AI search engines and how you should answer them. So the next step, what you have to do is essentially use this data that we just collected through this prompt, which is like use cases, benefits, negatives. Use your product specifications so that it's actually accurate, and then use a detailed prompt. So here I'm just going to go in a little more detail about prompts. You are to act as an expert in content strategy and product information optimization. Your task is to generate content using a question-answer optimization Q&A enhancement framework. This framework structures product information around natural customer queries, creating a conversational discovery path. The framework showed one question mapping, and then I say exactly how you should group questions, how to identify common customer questions, what is the question hierarchy even, because again, you can't put The Zillion Q&A is on your page. So you want to understand, okay, which is the most important question that I have to have on the page. And then I give it an answer architecture, you know, lead with direct response using key noun phrases, provide reporting evidence such context, include specific use cases, link to related features and benefits. Again, this is exactly the kind of shit that AI engines really love. Like, you know, you basically answer what normally people would love to find out. Then we have number three is question feature alignment, map questions to specific features, connect features to benefits, include real world applications. And then I give it actual examples so that I can make it like I can guide it even more as to like how my questions should be structured and also how long my and how my answers should be written. So this is a pretty neat prompt to like help you build this Q&A framework. Next up. All right. Well, guess what? It does it really well. It creates me a little table for each type of category. So like, you know, here I'm actually analyzing Stanley Tumblr. And, you know, a lot of people are talking about how much they love the portability of it, the fact that it keeps the drinks hot or cold. And here I actually like all of these questions based on a priority level, based on how frequently it was mentioned and with the answers. If you want to be in touch with me, I am on LinkedIn. Type in the long answer name. I'm pretty much the only one. You can subscribe to my AI for e-commerce newsletter. Basically, I write about AI and e-commerce three times a week. And I know that sounds quite a lot, but I regularly get people saying, hey, this is the only newsletter that I read and I read it in In its entirety, it just means no boring. It's funny. So that makes it easier. And yeah, I am very passionate about it. And then also I'm starting something new, which is a WhatsApp group where I can share the latest SOPs and just generally the things that I experiment. So if you want to learn in a very practical way, and you know, like here for me, like, you can scan that QR code and That's all folks. Speaker 2: Thank you so much for joining us. I'm so glad you were able to come on and share a little bit. I definitely have learned quite a bit. I know the folks here have too. I see we have comments. Thanks for a great session. That was great. Thank you. Thank you for the great presentation. I always learn something new. So it is much appreciated and we look forward to having you on again at some point. Speaker 1: Thank you. Thank you for having me.