Amazon Listing Secrets: How to Make Amazon’s Rufus Recommend YOUR Product

/
July 15, 2025
Share Post

In this episode, we welcome Joanna Lambadjieva, a leading authority in AI-powered e-commerce strategy and founder of Amazing Wave, a company that specializes in advanced Amazon listing optimization. Joanna has worked with top brands across Europe and North America, helping them channel the latest in artificial intelligence to transform their product listings, boost conversions, and stay ahead of Amazon’s algorithms. In this conversation, Joanna shares her expertise on leveraging customer data, unconventional research sources like Reddit, and the earth-moving impact of Amazon’s AI shopping assistant, Rufus. Read the full transcript below.

Episode 37 of The Seller’s Edge – Joanna and Jonathan talk about:

  • [00:00] AI In Ecommerce
  • [00:55] AI Has Transformed Listing Optimization
  • [03:07] How Does AI Understand Listings
  • [04:47] Leveraging Reddit
  • [05:31] Aggregating Insights with GummySearch
  • [07:01] Backend Optimization and Keywords
  • [09:40] Leveraging AI for Keyword Optimization
  • [10:31] Titles Should Be Written By Humans
  • [11:08] Discussing Rufus Patent
  • [11:58] The Key To Optimizing Titles
  • [12:50] Brandon Young’s Optimization Framework
  • [14:28] Highlighting Benefits In The Title
  • [16:24] Appealing To Specific Demographics
  • [19:27] Disconnect Between Listing Elements
  • [22:14] Pain Points For Implementing AI
  • [24:30] Ethical Concerns
  • [26:07] How AI Can Hurt Brand Reputation
  • [27:45] What’s Next For AI In Ecommerce?
  • [30:44] Recap and Closing Remarks

Key Takeaways:

  1. Leverage Diverse Data Sources: Use AI to analyze customer reviews, personas, and social platforms (like Reddit) for deeper insights into buyer motivations and pain points.
  2. Optimize for AI Gatekeepers: Structure listings to be contextually relevant, anticipating how AI systems like Rufus interpret and recommend products.
  3. Focus on Attributes and Context: Go beyond keywords to ensure backend attributes and contextual information are comprehensive and accurate.
  4. Personalize by Niche: Identify and target your primary customer persona rather than trying to appeal to everyone.
  5. Invest in AI Learning: Don’t give up after initial setbacks. Learn how to properly prompt and utilize AI tools for listing optimization and process automation.
  6. Maintain Authenticity and Ethics: Be transparent about AI-generated content and avoid misleading customers with synthetic UGC or manipulated imagery..

Full Transcript of Episode:

JONATHAN: How do you feel that AI has transformed product listing optimization in the last few years? 

JOANNA LAMBADJIEVA: Well, let’s say, predominantly in the last two years it’s changed a lot about two things. Firstly, how much data people can use in terms of informing their listings. So you know, before the predominant way to optimize a listing is you have a copywriter optimization person who sits down, gets some keywords, gets some, hopefully benefits, and then puts them all together, writes something and off we go. Now with AI, what we have seen is that you can actually get so many more rich data points like customer Personas, analyze reviews to get really the benefits, the use cases, the love and hate from your own customers, and other external data. Like, you know, for example, what I do with a lot of clients is look at Reddit and getting a lot of insight from there and then putting all of this data, all together into an AI large language model like ChatGPT or Claude, analyzing that and then producing one superior listing for a particular customer Persona type with the benefits that actually are resonating with that customer. So that’s one like you can absolutely use so much more data. The second thing is I think listings are not going to be any more just, you know, what we imagine or what we need them to be. It will be satisfying. An AI essentially algorithm like that then talks with our customer. So there is almost like an additional step. And you know what I’m talking about here, I’m talking about Rufus. And so I think what is going to happen and like what is happening right now is now we are starting to think about listings that are not just satisfying the customer, but first satisfying the AI chatbot that has to recommend them to the customer. And that is kind of like a much more contextual way to think about listings and not just keywords plus benefits equals a good optimized listing. 

JONATHAN: I love that. So do you think that that’s like the first kind of, to get past that, that gatekeeper of Rufus is to kind of take a listing, feed it into an AI and be like, hey, tell me what you think of this. And kind of like what, like, compared to other products on the market? I’m curious, like if you’ve, if you use it sort of as a translator, how do you gauge it from that perspective? 

JOANNA LAMBADJIEVA: It’s interesting, like, this was kind of the early school of thought in terms of how to deal with Rufus? No, I think it’s actually a little more complicated than that. I think at the end Rufus is just trying to satisfy the customer in a very natural way with the signals that it already knows about this customer and the signals that the customer is giving Rufus at the point of chat. And so in the end it’s not about, of course it’s about satisfying Rufus, but it’s about understanding what Rufus is looking for. And that is contextually relevant information that matches the intent of the customer. And so from that point of view, your first I guess step to optimizing for that is really, really pinpointing what is like who is your customer, what do they use your product for, what are the problems, aspirations, different use cases that they use your product for or your competitors products and then mapping that all around your listings so that actually when Rufus starts looking for all of these use cases amongst its many, many listings, it can match that information within your listing and then essentially swim out your result, in its answer. 

JONATHAN: I like that, I love that you call that Reddit too, because you are one of, I think there was only one other guest that called out Reddit as a place to really leverage for customer insights. And I think Reddit is such an untapped resource because it’s something that people rarely look at and they absolutely should. It’s the first place I go to when I’m looking for a product I like. Yeah, like what’s your favorite vacuum or whatever it is. Yeah, so it’s amazing that not more people are leveraging that. 

JOANNA LAMBADJIEVA: Yeah, no, I think it’s, it’s becoming a little more prominent. But yeah, I agree. I think it’s because until now there wasn’t really an easy way to essentially aggregate the data and get insights quickly. But the cool thing is now there are a couple of solutions. So, like, one of the tools that I really love using is called Gummy Search. And what it does is in essence you can create an audience around the topic. So let’s say you are researching in depth, I don’t know, vegan supplements or dog supplements. And so you can create your own audience based on a keyword and then you basically curate the different Reddit threads that are relevant to this topic. And further to that you can like basically Gummy Search aggregates the data and like, essentially gives you almost like a ranking of I guess trending topics in this particular sub niche and you can pull that data and then further use it with, let’s say Claude or ChatGPT to analyze it and further mine for insight. So it’s super, super useful. And I think actually Reddit is already rolling out something of their own like that where basically they can aggregate the data. Because I think, you know, more and more companies are realizing that, you know, the future of, whatever optimization and customer engagement we’re doing, we have to happen or start with the understanding of who the customer is and what their pain points are. And Reddit is just amazing for that. 

JONATHAN: Yeah, I mean you brought. There’s so many things I want to get into and it’s such an interesting topic and a good place to start is there are two things that happened in 2024. One was obviously the rollout of Rufus. Amazon also expanded the backend keywords from like 250 characters to, I think it’s 2,500 characters. So there’s a part of me being like, all right, well, like, Rufus is going to be really good at these, you know, behavioral, conceptual things. And it seems like it’s going to be less reliant on keywords. Like, it’ll be looking at things in reviews and questions that people are asking. So I’m curious, like, what does that mean for keywords? Like how, how much, how much, how much value should we be putting in keywords? And what are your thoughts as far as the back end being expanded to 2,500 characters and how valuable that is? Because I’ve always wondered if the backend keywords even really matter all that much. And then Amazon did this and I’m like, what do they know that I don’t? 

JOANNA LAMBADJIEVA: Obviously more so I think keywords are not gone right now, but I think optimization with Rufus in mind absolutely should be a priority. In terms of the back end, I like, you know, I’m definitely not an expert in backend. I think attributes matter more than anything now. I think it’s about really including as much as possible information on the back end and helping that sort of being fed into Rufus into like, essentially like mapping out everything to do with attributes around your product. Because this is the information that like Rufus essentially pulls, the backend keyword search term field. I don’t know. I think for several years I haven’t particularly given it any love. But that’s not to say that probably there might be some like, you know, juice that you might get out of it. I, I can’t, I can’t say yay or nay, but I think, you know, you just have to think about the future. And the future in my view is that we will have less and less say in terms of kind of keyword and optimization for listings. It will be more and more done by AI. It will be essentially the way I kind of imagine it in my head is in the future you upload your product and the AI and it already kind of does that. It scans your images, looks at your attributes and then kind of decides what is your product and then essentially is able to through your reviews, through your images and so forth, all of that information that you haven’t essentially typed in. It gives a lot more contextual data to the customer and then decides what to recommend. So yeah, I think there are keywords still, but not for that long I think. 

JONATHAN: Okay, I mean so while keywords are mattering, have you been leveraging AI for optimizing keywords or doing keyword research or anything like that? And can you talk a little bit about that? 

JOANNA LAMBADJIEVA: Sure. So I absolutely still use AI to do classic optimization. So it’s just that it becomes a much faster exercise. So like you know what I talked to you about in the first question, which is, you know all of the data input, so is your customer Persona, what do they love? What are they, what are their loves for the product? What are the benefits they’re looking for? What are the negatives that they hate about your competitor’s product or your product? All of that is super essential as the core of what you want to include in your listing. But then at the end you kind of do the master keyword stroke with AI helping you to sort of disseminate the keywords around your listing. Now the only exception that I would say is the title, which I am still obsessively control freakish about which I think is still a human’s job to do. Don’t let AI fudge with your title but you know, with sort of the much more complex Rufus paradigm of optimizing that, that is also something that will probably eventually end up with AI. One like tip for the listeners, SL viewers is I’m sure that if everyone has seen it but Oana, Danny and Andrew Bell have done a spectacular job of writing the Rufus blueprint. It’s a beast of a work. It’s 8,000 words on the Rufus patent and they break down all of the most important things that you need to know about what Rufus is taking. Like considering when recommending a product when analyzing your listing and then sort of giving you really actionable steps of how to optimize your listing. So for anyone who’s interested in how they should approach that in the future, this is, this is a must read. 

JONATHAN: Love it, love that recommendation. And then I mean I’m intrigued by your control freakishness about the title. What, what, can you share what your specific approaches with titles or how you think most people should be approaching titles? And it’s something that I feel like every guru on YouTube has different feelings on. So I’m curious if you have thoughts of a very specific approach. 

JOANNA LAMBADJIEVA: Firstly, I want to make the disclosure that I am not a guru so I do not, I do not fall in this category. I’m just a curious learner. So in terms of title, I, again, have been changing my mind a lot about it. I think the classic way is to understand. Okay. So you know, it really depends on where you are in terms of your, your journey, as a product, are you just launching, are you a mature product? You know where, what are you trying to achieve in the long term, short term. But yeah, it’s for me like I am a big fan of Brandon Young’s kind of optimization for framework, you know, with compos and root keywords. So it’s just about understanding. Okay, so in the short term where am I trying to move the dial? Like you know, which are my long low hanging fruit keywords that I can really attack by being very intentional with exact matches which are sort of like several root keywords that I can hit in like one go but then also not keyword stuffing and just including benefits. And I think again like going back to what I was talking about Rufus, it is about you know, using keywords sensibly but also going back to again those benefits that customers are looking for. And one of the interesting updates that you know Amazon announced at Accelerate last year was about the fact that they are now going to start testing or they already started testing like customizable titles where Amazon essentially changes or swaps the basically where it puts the different parts of your title like the, sorry the order of the keywords. So it can move around, you know, whether for example it would put the size right at the front of the title or the back based on the individual customer and what their preferences are. So it will be really interesting how that will develop and how that will influence titles. So again I think with titles we’ll lose more and more control over what and how we construct that. 

JONATHAN: Fascinating. But as far as kind of incorporating more like benefit information in the title, do you think that that’s something that people should stay away from or do you think that there are specific products where it’s good for? 

JOANNA LAMBADJIEVA: No, I think, I think benefits are super key. I mean again, like it, it really like there’s so many different products on Amazon. I can’t say that this is working for 100% of those. But you know, one of the things that I love doing and for me is like super eye opening is analysis with using Claude and reviews. So Claude is really, really smart in terms of semantic meaning. And so what I do is I like to pull a lot of reviews from a particular product and I essentially do quite a detailed prompt to try and understand what are the top, let’s say, benefits that customers are getting from this product. And what is cool about Claude is that instead of just counting the frequency of a word in a review, it counts the meaning of a particular, let’s say, phrase. So you know, and I’ve done this, this comparison with ChatGPT and Claude and ChatGPT like, you know, counts the number of times that like, love or like or you know, a word has been mentioned while Claude essentially analyzes groups of keyword keywords and their meaning and then counts that meaning how many times it, it was, it was mentioned. And so what’s amazing about it is I can figure out which is the top, most like, most important benefit that customers like mention again and again and again in reviews. And so if I have the top three most important things that customers are buying this product for, well, of course I’m going to try and put that into the title because I know that this is why most people go for that product. 

JONATHAN: I’m curious. One of the things that I’ve always, I guess not always because it’s a newer idea, but I mean personalization is becoming more and more, important and it’s a priority in marketing and a bigger trend and obviously that’s where a lot of things are going when it comes to a product listing. It’s a lot, it’s a lot more difficult. Obviously we’ll get to a future where you can like, you know, personalize things to IP addresses or something, but for now you have a generic listing that everybody looks at. Are there ways that you’ve thought of or approaches that you have to, to try to make things more personal to, or try to appeal to different demographics. Like maybe you designate a specific bullet to a product, bullet to one demographic, to another, to another or. Yeah, I’m just curious what your thoughts are there. 

JOANNA LAMBADJIEVA: Yeah, well, this is the thing like you know, again it goes back to non-AI stuff and you know, just basic, you know, marketing which is if you try to talk to too many, you speak to none. Right? So you have one customer Persona that is probably over index, mixing above everyone else. If you have like several then that’s cool. Then you know, you probably can do, you know, invest your time in, in making it broader. But probably there will be one or maximum two like types of people that your product really, really talks to. And so instead of trying to, you know, sell it to everyone, try to really nail that niche. Because the problem with Amazon and well, you know, most digital products, but Amazon particularly is that the competition is so fierce and like there is so much commoditization in so many categories that you know, everybody’s just trying to, you know, fudge the other by like, you know, copying, doing exactly the same thing. So if you know what your customer is, you know, responding to what is important for them, like you know, go for that. That is the best personal personalization you can do right now. And again, just read sort of like circling back to the previous answer. You know, you don’t have to guess like you have, you can, you can use AI to help you with all of this essentially audience work. So you know, you have some. If you are a seller in the US you already have the demographics tab in seller central. You can pull the data, as the. But even if you don’t have that, if you’re like you’re just starting or you are based in Europe and you don’t have that data, you can use Claude or ChatGPT or even perplexity to research, like who this, who most likely is that audience. And then from there you basically develop, you know, what it, what, like what is your, what are they like, you know, what are they important? What are the important things for them? What is the tone of voice that you have to develop and so forth and so on. One of the really cool things that I’m really looking forward to is that ChatGPT is going to roll out, maybe in the next few months is the wider availability of deep search. So deep search for the people who don’t geek out like me all the time About AI is a feature which essentially you give us you can give a complex prompt, but you can give a simple prompt which is give me a full research on I don’t know, like under eye bags and then is going to give you is a 200 page like super deep in like complex research on everything that is to do with under eye bags, including like analysis of like you know, YouTube and social media and so forth and so on. So like you know, if you want to really, you know understand the customer that buys eye cream for under eye bags of people of the age above 50, this is like an easy peasy way to do it. And I mean it’s currently available, you just have to drop $200 per month. But if you wait for a few months, like I think ChatGPT is going to give us some of that goodness for the plus plus, subscribers too. 

JONATHAN: I asked the question about the bullets and as far as trying to appeal to a different demographic because one of the things I notice a lot is a disconnect between bullet points in a product and I’m just like, why is this so vastly different than the bullet before it or after it or from all the other bullets? And then the other thing I notice is the disconnect between the bullet points on the product and then what’s in the image stack and the copy there. And I’m curious, is that something you’ve noticed and what is your approach or your thoughts on the relationship between copy in the image stack and the bullet points? 

JOANNA LAMBADJIEVA: Yeah, I mean absolutely. This is the thing, I think the world of Amazon sellers, I think they are such capable people because most of the profile of an Amazon seller is someone who has started this kind of on the side as a side hustle and then they build a multi million business. And you know, it’s like, yeah, whatever, I did that while also having a 9 to 5 job. Like kudos to you. They are so inventive and so great at figuring out solutions around the corner. But the one thing I’d say is that sometimes for a lot of sellers there is an obsession with hacks. We’re trying to figure out how to go around the system or figure out the next greatest thing which is some sort of like super smart tactic that nobody has ever heard that is going to like, you know, you know, fudge one over above their competitors. But they don’t go back to the basics which is, you know, who’s your customer? Is there a brand like, you know, are you talking to that customer? I’ve also noticed it with like, you know, proper, proper brand. Sometimes I look at the listing and they have only one customer Persona and then there is a picture of a completely different person. Like, because this is, this is some, something someone probably has copy pasted from another listing of another competitor. And so sometimes even just the common sense of just sit down and like architecture your own listing for that one person, it just doesn’t happen because these people are so busy, they have to do so many things and it’s just about, okay, let’s just get the volume, get all of the stuff done. But I think some really essential stuff sometimes gets missed out. It’s not all sellers, but I think there is an obsession with efficiency and sometimes important things and logical stuff get missed out. 

JONATHAN: 100%. And it’s not something I definitely don’t notice with all sellers. I mean there are those bullet points that I know that Amazon throws in there themselves and I’m just like. That makes that product listing rather ugly. But I’m curious, I mean we talked about the need for efficiency. In your experience seeing businesses try to use AI to optimize, what do you think have been sort of pain points or difficult aspects of it for them to implement? 

JOANNA LAMBADJIEVA: Yeah, I think it’s sort of the misconception that AI is just ChatGPT and AI is, you know, you, you put one little prompt and oh, look, it doesn’t work. Oh, it’s. AI is just a little like, it’s not working for me. And it’s, it’s just this kind of preconception that you don’t have to sit down and learn AI because it’s just really easy peasy shit, right? You just have to chat. So you, know that for me that, that is the big stumbling block is the understanding that just like, you know, learning how to optimize your advertising on Amazon or Facebook or whatever, you, there is still a lot that you have to learn about how to use these chatbots. What are their limitations, what they can and can do, how to structure, prompt, how to feed in the data that you want to be analyzed, or how to like, you know, break down a process in like composite parts that make sense to the large language model that you’re feeding in, which is the, like the correct large language model to use at a different task. So the stumbling blocks is that people get bad outputs and then they give up because they think other technology is not there. And yeah, sure, There are certain uses that maybe the technology is still not completely there, but actually I think for so many things there is an AI that can help them and optimize their process and make it faster and cheaper. And like, you know, this is one of the things that I help a lot of companies and agencies do is redesign their processes so that actually they don’t go through the manual process again and again and again. 

JONATHAN: It’s definitely a mistake that I’ve seen people make, especially with just feeding prompts. And I’m just like, don’t talk to it like that. Like there’s like, there is like a proper etiquette that you need to follow, follow when talking to generative AI. So that’s a really good call out. As far as ethical concerns, what do you think businesses should be keeping in mind? 

JOANNA LAMBADJIEVA: Okay, so if, if I, if I just narrow it down, I’m not going to go super wide because I think there’s enough people talking about that. But I think if I just narrow it down just in the sort of the little bubble of eCommerce marketing, I think people need to realize that, you know, you are still talking and selling to people and you should use AI to create marketing copy and images and videos and whatnot. But people are not stupid. And if you like, if we always marketers keep pretending that like you know, what we are showing, for example in terms of imagery or video or what people now are raving about, which is UGC AI. We are not upfront that this is fake stuff. We are shooting ourselves massively in the foot because then it will be so much more difficult to earn back that consumer trust. So from an ethical point of view, I think it’s just about you know, being really upfront with your customers. I guarantee you, every time, more and more, mid journey images will come out–and I also teach how to work with it. But the more and more these AI images come out, the more and more people will kind of figure out, okay, this is AI generated and yes, at some point they’ll become really seamless. But should you just always rinse the AI images? Maybe not. Or if you are upfront with your customers. I am particularly allergic to the AI UGC. I think there are definitely use cases for doing AI avatars to optimize some workflows. But to pretend that somebody is recommending a product that is an actual AI avatar, making it seem like this is social media content I think is just, it’s not okay. And I think people shouldn’t do it. I just think like, you know, we, we are marketers, but we are also consumers. Do we want to be manipulated? And like, do we want to be convinced by things that are not real? Like, you know, and, and this will become much more prevalent. Like, you know, what is real? Who are we, the people we trust online? Who are the brands that we trust online? Like, this is going to become much more an important topic. And I think the brands who are genuine and like, yes, they use AI, but in a responsible way, in a transparent way, I think they are going to be the one who. But if you try to cheat your customer, it’s going to come back. 

JONATHAN: I really, really appreciate that perspective because it’s about maintaining authenticity even if you’re using artificial intelligence. And I think that that’s really important and that’s just a cornerstone of marketing in general. Like, I don’t think you’re ever going to win if you don’t try to maintain some sense of authenticity. So that’s a really great call out and you sort of answered this, in your last question. But I’m curious if there’s other elements to it, like how do you see this evolving? Like, where do you see this going and the ability to utilize it for, I mean one on the consumer side, like something like Rufus, but also on, you know, the seller side and how we can use these tools in the future and what they’ll be capable of. 

JOANNA LAMBADJIEVA: I am both equally excited and also worried about the agentic kind of AI that is coming. And it’s coming very soon. I think, you know, what people are talking about. Agentic right now are not really agentic, it’s just automations. Agentic is what OpenAI is talking about when it talks about Operator and what Claude rolled out in a very mini version like several months ago. It’s basically when you give a task to a large language model and then it basically decides what are the next steps to achieve that task and then achieves it. I’m really excited because obviously this is really like having an intern that is at your fingertips, which is awesome. It’s a cliche, but it’s good. It’s going to be awesome. But on the other side, I’m really worried about how many jobs this is going to destroy. Would the world be able to create enough jobs to cover the ones that it loses? Is there going to be a bigger disparity between people? Because there already is a really large disparity and this is going to create even more of a gap perhaps. So yeah, so I think from a work point of view, yay. But from an over macro point of view, like you know it could, it could be very difficult for a lot of people for a certain amount of time. So that’s one. I think from a consumer point of view again it will be people will get totally turned off. My prediction of, of, I don’t know, online I feel like you know we are right now spewing content at the rate of, I don’t know, like light, speed of light. And so I think the Internet is going to become really very black in the next two years and so just like Gen Z, I read an article that says the sales of stupid phones, like you know the Nokias of this world have shot up in recent years. I think this is going to be just everyone. I think like you know the Internet will be just a place of just fakeness and shit content and I think there will be some like little islands which hopefully will, will provide like human generated content at a quality but most of it will be clever. And so I think there will be a really big turn off point for a lot of people online which is going to be very interesting for us eCommerce sellers and, and marketers, how we’re going to tackle that. I think it will also be brilliant. From an education point of view. I think like you know one of the things that I’ve noticed is how much I have upskilled myself by using AI and I think it would be awesome if more people use it this way. You know my best friend, she works for an NGO and she’s now looking for ways to essentially develop projects with AI to help underrepresented communities in Africa which is going to be so awesome. So I really hope that AI is also going to be used in that sort of context. But yeah like I think it’s, I have a very mixed view towards the future. I hope it’s going to be positive. I’m not really sure. 

JONATHAN: I think that’s a healthy perspective to have on it and I appreciate that. I do agree with you that there’s going to be a lot of like content out there. How people react to it is the thing that I’m waiting to see how, how that plays out. Hopefully humanity proves my theory that they are deep down good. Right. Rather than otherwise. Yeah. Joe, this has been a really great conversation. You’ve given us a lot of really insightful perspectives and actionable items, too, and advice, which I really love. And I do enjoy your sense of humor as well and your humanistic approach to things. This has been a really great conversation.

JONATHAN: How do you feel that AI has transformed product listing optimization in the last few years? 

JOANNA LAMBADJIEVA: Well, let’s say, predominantly in the last two years it’s changed a lot about two things. Firstly, how much data people can use in terms of informing their listings. So you know, before the predominant way to optimize a listing is you have a copywriter optimization person who sits down, gets some keywords, gets some, hopefully benefits, and then puts them all together, writes something and off we go. Now with AI, what we have seen is that you can actually get so many more rich data points like customer Personas, analyze reviews to get really the benefits, the use cases, the love and hate from your own customers, and other external data. Like, you know, for example, what I do with a lot of clients is look at Reddit and getting a lot of insight from there and then putting all of this data, all together into an AI large language model like ChatGPT or Claude, analyzing that and then producing one superior listing for a particular customer Persona type with the benefits that actually are resonating with that customer. So that’s one like you can absolutely use so much more data. The second thing is I think listings are not going to be any more just, you know, what we imagine or what we need them to be. It will be satisfying. An AI essentially algorithm like that then talks with our customer. So there is almost like an additional step. And you know what I’m talking about here, I’m talking about Rufus. So, you know, this is right now still kind of like at the beginning, right? So not so many people have adopted using Rufus when they research, but Amazon is pushing it really aggressively. And so I think what is going to happen and like what is happening right now is now we are starting to think about listings that are not just satisfying the customer, but first satisfying the AI chatbot that has to recommend them to the customer. And that is kind of like a much more contextual way to think about listings and not just keywords plus benefits equals a good optimized listing. 

JONATHAN: I love that. So do you think that that’s like the first kind of, to get past that, that gatekeeper of Rufus is to kind of take a listing, feed it into an AI and be like, hey, tell me what you think of this. And kind of like what, like, compared to other products on the market? I’m curious, like if you’ve, if you use it sort of as a translator, how do you gauge it from that perspective? 

JOANNA LAMBADJIEVA: It’s interesting, like, this was kind of the early school of thought in terms of how to deal with Rufus? No, I think it’s actually a little more complicated than that. I think at the end Rufus is just trying to satisfy the customer in a very natural way with the signals that it already knows about this customer and the signals that the customer is giving Rufus at the point of chat. And so in the end it’s not about, of course it’s about satisfying Rufus, but it’s about understanding what Rufus is looking for. And that is contextually relevant information that matches the intent of the customer. And so from that point of view, your first I guess step to optimizing for that is really, really pinpointing what is like who is your customer, what do they use your product for, what are the problems, aspirations, different use cases that they use your product for or your competitors products and then mapping that all around your listings so that actually when Rufus starts looking for all of these use cases amongst its many, many listings, it can match that information within your listing and then essentially swim out your result, in its answer. 

JONATHAN: I like that, I love that you call that Reddit too, because you are one of, I think there was only one other guest that called out Reddit as a place to really leverage for customer insights. And I think Reddit is such an untapped resource because it’s something that people rarely look at and they absolutely should. It’s the first place I go to when I’m looking for a product I like. Yeah, like what’s your favorite vacuum or whatever it is. Yeah, so it’s amazing that not more people are leveraging that. 

JOANNA LAMBADJIEVA: Yeah, no, I think it’s, it’s becoming a little more prominent. But yeah, I agree. I think it’s because until now there wasn’t really an easy way to essentially aggregate the data and get insights quickly. But the cool thing is now there are a couple of solutions. So, like, one of the tools that I really love using is called Gummy Search. And what it does is in essence you can create an audience around the topic. So let’s say you are researching in depth, I don’t know, vegan supplements or dog supplements. And so you can create your own audience based on a keyword and then you basically curate the different Reddit threads that are relevant to this topic. And further to that you can like basically Gummy Search aggregates the data and like, essentially gives you almost like a ranking of I guess trending topics in this particular sub niche and you can pull that data and then further use it with, let’s say Claude or ChatGPT to analyze it and further mine for insight. So it’s super, super useful. And I think actually Reddit is already rolling out something of their own like that where basically they can aggregate the data. Because I think, you know, more and more companies are realizing that, you know, the future of, whatever optimization and customer engagement we’re doing, we have to happen or start with the understanding of who the customer is and what their pain points are. And Reddit is just amazing for that. 

JONATHAN: Yeah, I mean you brought. There’s so many things I want to get into and it’s such an interesting topic and a good place to start is there are two things that happened in 2024. One was obviously the rollout of Rufus. Amazon also expanded the backend keywords from like 250 characters to, I think it’s 2,500 characters. So there’s a part of me being like, all right, well, like, Rufus is going to be really good at these, you know, behavioral, conceptual things. And it seems like it’s going to be less reliant on keywords. Like, it’ll be looking at things in reviews and questions that people are asking. So I’m curious, like, what does that mean for keywords? Like how, how much, how much, how much value should we be putting in keywords? And what are your thoughts as far as the back end being expanded to 2,500 characters and how valuable that is? Because I’ve always wondered if the backend keywords even really matter all that much. And then Amazon did this and I’m like, what do they know that I don’t? 

JOANNA LAMBADJIEVA: Obviously more so I think keywords are not gone right now, but I think optimization with Rufus in mind absolutely should be a priority. In terms of the back end, I like, you know, I’m definitely not an expert in backend. I think attributes matter more than anything now. I think it’s about really including as much as possible information on the back end and helping that sort of being fed into Rufus into like, essentially like mapping out everything to do with attributes around your product. Because this is the information that like Rufus essentially pulls, the backend keyword search term field. I don’t know. I think for several years I haven’t particularly given it any love. But that’s not to say that probably there might be some like, you know, juice that you might get out of it. I, I can’t, I can’t say yay or nay, but I think, you know, you just have to think about the future. And the future in my view is that we will have less and less say in terms of kind of keyword and optimization for listings. It will be more and more done by AI. It will be essentially the way I kind of imagine it in my head is in the future you upload your product and the AI and it already kind of does that. It scans your images, looks at your attributes and then kind of decides what is your product and then essentially is able to through your reviews, through your images and so forth, all of that information that you haven’t essentially typed in. It gives a lot more contextual data to the customer and then decides what to recommend. So yeah, I think there are keywords still, but not for that long I think. 

JONATHAN: Okay, I mean so while keywords are mattering, have you been leveraging AI for optimizing keywords or doing keyword research or anything like that? And can you talk a little bit about that? 

JOANNA LAMBADJIEVA: Sure. So I absolutely still use AI to do classic optimization. So it’s just that it becomes a much faster exercise. So like you know what I talked to you about in the first question, which is, you know all of the data input, so is your customer Persona, what do they love? What are they, what are their loves for the product? What are the benefits they’re looking for? What are the negatives that they hate about your competitor’s product or your product? All of that is super essential as the core of what you want to include in your listing. But then at the end you kind of do the master keyword stroke with AI helping you to sort of disseminate the keywords around your listing. Now the only exception that I would say is the title, which I am still obsessively control freakish about which I think is still a human’s job to do. Don’t let AI fudge with your title but you know, with sort of the much more complex Rufus paradigm of optimizing that, that is also something that will probably eventually end up with AI. One like tip for the listeners, SL viewers is I’m sure that if everyone has seen it but Oana, Danny and Andrew Bell have done a spectacular job of writing the Rufus blueprint. It’s a beast of a work. It’s 8,000 words on the Rufus patent and they break down all of the most important things that you need to know about what Rufus is taking. Like considering when recommending a product when analyzing your listing and then sort of giving you really actionable steps of how to optimize your listing. So for anyone who’s interested in how they should approach that in the future, this is, this is a must read. 

JONATHAN: Love it, love that recommendation. And then I mean I’m intrigued by your control freakishness about the title. What, what, can you share what your specific approaches with titles or how you think most people should be approaching titles? And it’s something that I feel like every guru on YouTube has different feelings on. So I’m curious if you have thoughts of a very specific approach. 

JOANNA LAMBADJIEVA: Firstly, I want to make the disclosure that I am not a guru so I do not, I do not fall in this category. I’m just a curious learner. So in terms of title, I, again, have been changing my mind a lot about it. I think the classic way is to understand. Okay. So you know, it really depends on where you are in terms of your, your journey, as a product, are you just launching, are you a mature product? You know where, what are you trying to achieve in the long term, short term. But yeah, it’s for me like I am a big fan of Brandon Young’s kind of optimization for framework, you know, with compos and root keywords. So it’s just about understanding. Okay, so in the short term where am I trying to move the dial? Like you know, which are my long low hanging fruit keywords that I can really attack by being very intentional with exact matches which are sort of like several root keywords that I can hit in like one go but then also not keyword stuffing and just including benefits. And I think again like going back to what I was talking about Rufus, it is about you know, using keywords sensibly but also going back to again those benefits that customers are looking for. And one of the interesting updates that you know Amazon announced at Accelerate last year was about the fact that they are now going to start testing or they already started testing like customizable titles where Amazon essentially changes or swaps the basically where it puts the different parts of your title like the, sorry the order of the keywords. So it can move around, you know, whether for example it would put the size right at the front of the title or the back based on the individual customer and what their preferences are. So it will be really interesting how that will develop and how that will influence titles. So again I think with titles we’ll lose more and more control over what and how we construct that. 

JONATHAN: Fascinating. But as far as kind of incorporating more like benefit information in the title, do you think that that’s something that people should stay away from or do you think that there are specific products where it’s good for? 

JOANNA LAMBADJIEVA: No, I think, I think benefits are super key. I mean again, like it, it really like there’s so many different products on Amazon. I can’t say that this is working for 100% of those. But you know, one of the things that I love doing and for me is like super eye opening is analysis with using Claude and reviews. So Claude is really, really smart in terms of semantic meaning. And so what I do is I like to pull a lot of reviews from a particular product and I essentially do quite a detailed prompt to try and understand what are the top, let’s say, benefits that customers are getting from this product. And what is cool about Claude is that instead of just counting the frequency of a word in a review, it counts the meaning of a particular, let’s say, phrase. Claude essentially analyzes groups of keyword keywords and their meaning and then counts that meaning how many times it, it was, it was mentioned. And so what’s amazing about it is I can figure out which is the top, most like, most important benefit that customers like mention again and again and again in reviews. And so if I have the top three most important things that customers are buying this product for, well, of course I’m going to try and put that into the title because I know that this is why most people go for that product. 

JONATHAN: I’m curious. One of the things that I’ve always, I guess not always because it’s a newer idea, but I mean personalization is becoming more and more, important and it’s a priority in marketing and a bigger trend and obviously that’s where a lot of things are going when it comes to a product listing. It’s a lot, it’s a lot more difficult. Obviously we’ll get to a future where you can like, you know, personalize things to IP addresses or something, but for now you have a generic listing that everybody looks at. Are there ways that you’ve thought of or approaches that you have to, to try to make things more personal to, or try to appeal to different demographics. Like maybe you designate a specific bullet to a product, bullet to one demographic, to another, to another or. Yeah, I’m just curious what your thoughts are there. 

JOANNA LAMBADJIEVA: Yeah, well, this is the thing like you know, again it goes back to non-AI stuff and you know, just basic, you know, marketing which is if you try to talk to too many, you speak to none. Right? So you have one customer Persona that is probably over index, mixing above everyone else. If you have like several then that’s cool. Then you know, you probably can do, you know, invest your time in, in making it broader. But probably there will be one or maximum two like types of people that your product really, really talks to. And so instead of trying to, you know, sell it to everyone, try to really nail that niche. Because the problem with Amazon and well, you know, most digital products, but Amazon particularly is that the competition is so fierce and like there is so much commoditization in so many categories that you know, everybody’s just trying to, you know, fudge the other by like, you know, copying, doing exactly the same thing. So if you know what your customer is, you know, responding to what is important for them, like you know, go for that. That is the best personal personalization you can do right now. And again, just read sort of like circling back to the previous answer. You know, you don’t have to guess like you have, you can, you can use AI to help you with all of this essentially audience work. So you know, you have some. If you are a seller in the US you already have the demographics tab in seller central. You can pull the data, as the. But even if you don’t have that, if you’re like you’re just starting or you are based in Europe and you don’t have that data, you can use Claude or ChatGPT or even perplexity to research, like who this, who most likely is that audience. And then from there you basically develop, you know, what it, what, like what is your, what are they like, you know, what are they important? What are the important things for them? What is the tone of voice that you have to develop and so forth and so on. 

JONATHAN: I asked the question about the bullets and as far as trying to appeal to a different demographic because one of the things I notice a lot is a disconnect between bullet points in a product and I’m just like, why is this so vastly different than the bullet before it or after it or from all the other bullets? And then the other thing I notice is the disconnect between the bullet points on the product and then what’s in the image stack and the copy there. And I’m curious, is that something you’ve noticed and what is your approach or your thoughts on the relationship between copy in the image stack and the bullet points? 

JOANNA LAMBADJIEVA: Yeah, I mean absolutely. This is the thing, I think the world of Amazon sellers, I think they are such capable people because most of the profile of an Amazon seller is someone who has started this kind of on the side as a side hustle and then they build a multi million business. And you know, it’s like, yeah, whatever, I did that while also having a 9 to 5 job. Like kudos to you. They are so inventive and so great at figuring out solutions around the corner. But the one thing I’d say is that sometimes for a lot of sellers there is an obsession with hacks. We’re trying to figure out how to go around the system or figure out the next greatest thing which is some sort of like super smart tactic that nobody has ever heard that is going to like, you know, you know, fudge one over above their competitors. But they don’t go back to the basics which is, you know, who’s your customer? Is there a brand like, you know, are you talking to that customer? I’ve also noticed it with like, you know, proper, proper brand. Sometimes I look at the listing and they have only one customer Persona and then there is a picture of a completely different person. Like, because this is, this is some, something someone probably has copy pasted from another listing of another competitor. And so sometimes even just the common sense of just sit down and like architecture your own listing for that one person, it just doesn’t happen because these people are so busy, they have to do so many things and it’s just about, okay, let’s just get the volume, get all of the stuff done. But I think some really essential stuff sometimes gets missed out. It’s not all sellers, but I think there is an obsession with efficiency and sometimes important things and logical stuff get missed out. 

JONATHAN: 100%. And it’s not something I definitely don’t notice with all sellers. I mean there are those bullet points that I know that Amazon throws in there themselves and I’m just like. That makes that product listing rather ugly. But I’m curious, I mean we talked about the need for efficiency. In your experience seeing businesses try to use AI to optimize, what do you think have been sort of pain points or difficult aspects of it for them to implement? 

JOANNA LAMBADJIEVA: Yeah, I think it’s sort of the misconception that AI is just ChatGPT and AI is, you know, you, you put one little prompt and oh, look, it doesn’t work. Oh, it’s. AI is just a little like, it’s not working for me. And it’s, it’s just this kind of preconception that you don’t have to sit down and learn AI because it’s just really easy peasy shit, right? You just have to chat. So you, know that for me that, that is the big stumbling block is the understanding that just like, you know, learning how to optimize your advertising on Amazon or Facebook or whatever, you, there is still a lot that you have to learn about how to use these chatbots. What are their limitations, what they can and can do, how to structure, prompt, how to feed in the data that you want to be analyzed, or how to like, you know, break down a process in like composite parts that make sense to the large language model that you’re feeding in, which is the, like the correct large language model to use at a different task. So the stumbling blocks is that people get bad outputs and then they give up because they think other technology is not there. And yeah, sure, There are certain uses that maybe the technology is still not completely there, but actually I think for so many things there is an AI that can help them and optimize their process and make it faster and cheaper. And like, you know, this is one of the things that I help a lot of companies and agencies do is redesign their processes so that actually they don’t go through the manual process again and again and again. 

JONATHAN: It’s definitely a mistake that I’ve seen people make, especially with just feeding prompts. And I’m just like, don’t talk to it like that. Like there’s like, there is like a proper etiquette that you need to follow, follow when talking to generative AI. So that’s a really good call out. As far as ethical concerns, what do you think businesses should be keeping in mind? 

JOANNA LAMBADJIEVA: Okay, so if, if I, if I just narrow it down, I’m not going to go super wide because I think there’s enough people talking about that. But I think if I just narrow it down just in the sort of the little bubble of eCommerce marketing, I think people need to realize that, you know, you are still talking and selling to people and you should use AI to create marketing copy and images and videos and whatnot. But people are not stupid. And if you like, if we always marketers keep pretending that like you know, what we are showing, for example in terms of imagery or video or what people now are raving about, which is UGC AI. We are not upfront that this is fake stuff. We are shooting ourselves massively in the foot because then it will be so much more difficult to earn back that consumer trust. So from an ethical point of view, I think it’s just about you know, being really upfront with your customers. I guarantee you, every time, more and more, mid journey images will come out–and I also teach how to work with it. But the more and more these AI images come out, the more and more people will kind of figure out, okay, this is AI generated and yes, at some point they’ll become really seamless. But should you just always rinse the AI images? Maybe not. Or if you are upfront with your customers. I am particularly allergic to the AI UGC. I think there are definitely use cases for doing AI avatars to optimize some workflows. But to pretend that somebody is recommending a product that is an actual AI avatar, making it seem like this is social media content I think is just, it’s not okay. And I think people shouldn’t do it. I just think like, you know, we, we are marketers, but we are also consumers. Do we want to be manipulated? And like, do we want to be convinced by things that are not real? Like, you know, and, and this will become much more prevalent. Like, you know, what is real? Who are we, the people we trust online? Who are the brands that we trust online? Like, this is going to become much more an important topic. And I think the brands who are genuine and like, yes, they use AI, but in a responsible way, in a transparent way, I think they are going to be the one who. But if you try to cheat your customer, it’s going to come back. 

JONATHAN: I really, really appreciate that perspective because it’s about maintaining authenticity even if you’re using artificial intelligence. And I think that that’s really important and that’s just a cornerstone of marketing in general. Like, I don’t think you’re ever going to win if you don’t try to maintain some sense of authenticity. So that’s a really great call out and you sort of answered this, in your last question. But I’m curious if there’s other elements to it, like how do you see this evolving? Like, where do you see this going and the ability to utilize it for, I mean one on the consumer side, like something like Rufus, but also on, you know, the seller side and how we can use these tools in the future and what they’ll be capable of. 

JOANNA LAMBADJIEVA: I am both equally excited and also worried about the agentic kind of AI that is coming. And it’s coming very soon. I think, you know, what people are talking about. Agentic right now are not really agentic, it’s just automations. Agentic is what OpenAI is talking about when it talks about Operator and what Claude rolled out in a very mini version like several months ago. It’s basically when you give a task to a large language model and then it basically decides what are the next steps to achieve that task and then achieves it. I’m really excited because obviously this is really like having an intern that is at your fingertips, which is awesome. It’s a cliche, but it’s good. It’s going to be awesome. But on the other side, I’m really worried about how many jobs this is going to destroy. Would the world be able to create enough jobs to cover the ones that it loses? Is there going to be a bigger disparity between people? Because there already is a really large disparity and this is going to create even more of a gap perhaps. So yeah, so I think from a work point of view, yay. But from an over macro point of view, like you know it could, it could be very difficult for a lot of people for a certain amount of time. So that’s one. I think from a consumer point of view again it will be people will get totally turned off. My prediction of, of, I don’t know, online I feel like you know we are right now spewing content at the rate of, I don’t know, like light, speed of light. And so I think the Internet is going to become really very black in the next two years and so just like Gen Z, I read an article that says the sales of stupid phones, like you know the Nokias of this world have shot up in recent years. I think this is going to be just everyone. I think like you know the Internet will be just a place of just fakeness and shit content and I think there will be some like little islands which hopefully will, will provide like human generated content at a quality but most of it will be clever. And so I think there will be a really big turn off point for a lot of people online which is going to be very interesting for us eCommerce sellers and, and marketers, how we’re going to tackle that. I think it will also be brilliant. From an education point of view. I think like you know one of the things that I’ve noticed is how much I have upskilled myself by using AI and I think it would be awesome if more people use it this way. You know my best friend, she works for an NGO and she’s now looking for ways to essentially develop projects with AI to help underrepresented communities in Africa which is going to be so awesome. So I really hope that AI is also going to be used in that sort of context. But yeah like I think it’s, I have a very mixed view towards the future. I hope it’s going to be positive. I’m not really sure. 

JONATHAN: I think that’s a healthy perspective to have on it and I appreciate that. I do agree with you that there’s going to be a lot of like content out there. How people react to it is the thing that I’m waiting to see how, how that plays out. Hopefully humanity proves my theory that they are deep down good. Right. Rather than otherwise. Yeah. Joe, this has been a really great conversation. You’ve given us a lot of really insightful perspectives and actionable items, too, and advice, which I really love. And I do enjoy your sense of humor as well and your humanistic approach to things. This has been a really great conversation.

Want more insights? Check out more episodes of The Seller’s Edge Podcast.

The show has covered many topics including PPC strategies, International markets,  DTC tips, creative content, and how to get those 5-star reviews easily.

Also, don’t forget to follow Viral Launch on LinkedIn, Facebook, or Instagram, and subscribe to the podcast on Spotify, Apple Podcasts, Audible, Soundcloud, or any platform you prefer.

If you’re a new seller who is looking to 10X their business, here are some really valuable resources to get you started:

  • Viral Launch – Viral Launch offers the most powerful suite of seller tools on the market, designed to optimize listings, drive traffic, boost conversions, and maximize profits.
  • Chrome Extension – Get real-time Amazon data and competitor insights with the Market Intelligence Chrome Extension, helping you make informed decisions with strong research and analyses of the market.
  • Free Tools for Sellers to Leverage – Keeping the budget tight? We get that. In that case, feel free to access our free tools, such as the FBA Calculator, to streamline your business operations and make smarter, data-driven decisions for your brand.

Don’t miss the chance to make your Amazon business soar. Subscribe to The Seller’s Edge podcast and explore Viral Launch’s valuable array of services and tools today!

Try Viral Launch, the all-in-one solution for new and established Amazon sellers.

Set Your
Amazon Business
Up For Success

Never miss an update with Viral Launch!
Subscribe to our exclusive newsletter for the competitive edge you need. Get essential tips, trends, strategies, and the latest news to elevate your Amazon business.

YOU MAY ALSO LIKE

SEE HOW VIRAL LAUNCH CAN HELP GROW YOUR BUSINESS