S3 E5 Optimising Ecommerce Brands for AI and Amazon Rufus

AI shopping assistants like Amazon's Rufus are fundamentally changing how customers discover and buy products, shifting power from search to AI.

Andri Sadlak breaks down the rapid rise of agentic commerce, explaining why consumers increasingly trust AI tools over traditional marketplaces for purchase recommendations. We explore crucial data, including how shoppers using Rufus convert at a staggering 36%, and why small businesses are currently lagging in Generative Engine Optimisation (GEO) compared to enterprise brands. 

This discussion provides a foundational look at the future of digital storefronts and how sellers must adapt their visibility strategies for the new retail landscape. 

Key insights discussed:

- The evolution of the customer journey from traditional search to AI-assisted checkouts. 

- Why Amazon and other tech giants are heavily investing in integrating AI into the shopping experience. 

- Actionable perspectives on reviewing competitor prompts and optimising listings for large language models. 

TIMESTAMPS 
00:00 - Introduction to Andri Sadlak and his e-commerce background 
03:31 - The concept of agentic commerce 
05:46 - The shift toward AI product discovery and Amazon Rufus 
08:52 - What is agentic commerce and how does it work? 
11:45 - Shopify's AI integration and the rapid adoption of ChatGPT 
16:49 - Rufus conversion rates and Amazon's AI push
20:45 - Which businesses are adopting Generative Engine Optimisation? 
43:31 - Reviewing competitor prompts and adapting optimisation strategies 
01:09:14 - Final thoughts on the ever-evolving AI landscape

Transcript

SPEAKER_01 0:08

Hi everyone, welcome to another episode of the online hustle. As you see, today I have Andre with us. Andre, welcome.

SPEAKER_00 0:18

Thank you so much. Pleasure and be here.

SPEAKER_01 0:20

Great. Well, thank you so much for joining us. Um, I think we'll go ahead and start and tell our viewers who you are, where you came from. Your background is very, very interested. So um interesting. So I I want everyone to know the entire background story um from you.

SPEAKER_00 0:39

100%. Yeah, I've been selling uh for quite some time uh since 2017. Sold one of the brands, uh Kafala Software, and um also worked with an aggregator, one one of those who survived and has actually done pretty well. Uh so definitely learned a lot from that experience. And now I'm focused on Gen Ecommerce, so helping brands optimize for the AI visibility and the future of how we'll be shopping.

SPEAKER_01 1:08

Yeah, amazing. I think you've done it all. Um, and you've succeeded in everything you've done based. I've I followed you through your trajectory and you've succeeded through all of them. You've helped brands become seven figures, eight figures, um, and now you've incorporated AI. So the focus today, uh, what's it gonna be?

SPEAKER_00 1:28

Yeah, thanks. First of all, it's high praise coming from Minaton. Yeah, the focus is gonna be uh uh on agente commerce explaining what it means and what we can do already, uh, and maybe how we can prepare for the shift that uh I see is already on the way.

SPEAKER_01 1:44

Perfect. Um so Andre's uh prepared a couple of slides that we'll we'll show now so that it gives you a bit of an idea on what the shift is, give you a bit of uh visibility rather than us just talking. So we'll do that then.

SPEAKER_00 1:58

Let's do it. Uh we'll make it interactive. Um, but I'll go through them so that nothing gets missed. Uh, and we can obviously uh interact at any moment, ask questions, make it make it fun. And just before I dive into it, wanted to just give a little bit of a disclaimer. Uh, everything useful in this presentation comes from my experience with the clients, with my team, um, and with me working with uh other sellers alongside other sellers since 2017. All the bad ideas are probably mine. Uh so if you do see that something's wrong or you know better, obviously speak up, Natalia, and also anyone else, feel free to comment because at the end of the day, we'll just all leave smarter. And that's my goal. Uh, and I've definitely made my fair share of mistakes. I even have to sell my house at some point to save the business. So definitely take it with a grain of salt. Um, and I'm not here to reach, I'm here to brainstorm and share what I've learned so far. Yeah, and a couple of things just to outline them my goals. Whatever you take from this from this conversation today, I want to do two things. One, uh use it for good. So please don't weaponize it against your competitors or deceive your customers, your shoppers. And two, if you find it useful, pass it on, uh, pay it forward, share it with other ethical brand builders in the space. Okay?

SPEAKER_01 3:23

Okay. Love that. That's what this is for. We just want to share information, make sure that everyone learns from each other. So I love those two.

SPEAKER_00 3:31

We're gonna speak about Agenti Commerce and specifically three levers to boost your brand's AI visibility. My name is Andrew Sadlak, I'm a founding head of product and strategy at Azama. And here's the question: what if your next customer never searched for your product, never read your listing, and still bought from you? Sounds pretty cool, right?

SPEAKER_01 3:57

Yeah.

SPEAKER_00 3:58

Yeah, and my name's Andre. I've been a seller since 2017. I exited my brand in late 2020. I co-founded Product Pinion. I scaled seven and eight-figure brands uh with an aggregator and as a consultant, and now I'm building Azoma, which is a Gent e-commerce optimization software.

SPEAKER_01 4:15

See, so it wasn't long when I said that you've done it all.

SPEAKER_00 4:19

Thank you. So just for clarity, Azoma is a software tool that helps uh brands like Lipton, Mars, PNG, Inilever, Breasdorf, L'Oreal, Colgate, HP uh optimize for showing up more often in large language models, uh, and specifically on Amazon and Walmart. That's uh what we do. And my promise to you is by the end of this conversation, you leave with some practical action items you can actually implement or at least pass on to your team to implement. Because that's that's my um, like I just hate it when I listen to something and it's all great ideas, all amazing. We're like future is here, but there's nothing I can do, right? So I want to fix that. Uh, and um that's that's my promise to you that I'm gonna make it as practical as I possibly can in the conversation.

SPEAKER_01 5:06

Nice.

SPEAKER_00 5:07

So we'll I'll break it down into three parts. Uh we'll start with the general understanding, so the shift, what I call the shift, why AI now owns the shelf. Then I'll dive into specifics, so three Amazon levers, so specific to Amazon's agent e commerce, the science behind it, and your action plan, and then we'll open it up to QA, we'll brainstorm together. Well, I'll I'll I'll share everything I can when you ask questions, Natalia.

SPEAKER_01 5:33

Perfect.

SPEAKER_00 5:34

Let's go for it. Sweet. So the shift, why AI now owns the shell? So quick question to you do you use Chat GPT?

SPEAKER_01 5:45

Almost daily, and I think that would be everyone's answer.

SPEAKER_00 5:49

Probably by now, right? Now, follow up. Do you use ChatGPT to decide what products to buy?

SPEAKER_01 5:56

I don't personally, and maybe this is being too honest, but I I am I still use a traditional method. I look through Amazon, I look at reviews. Um, sometimes I guess if you need ideas on like a gift, if I want to buy someone a gift, I might put into Chat GPT what they like, um, what their hobbies are, get ideas on on what to buy them, and then I will shift onto the marketplace to look for that.

SPEAKER_00 6:25

Yes. So what you're describing is basically the awareness stage when you're looking for the solution to your problem. And what do you describe is basically the average person's behavior right now. Okay. Now here's another question. Do you shop on Amazon? Of course. Of course, right. Like if we all have. Right? Have you tried Rufus though to decide what products to buy?

SPEAKER_01 6:47

I personally haven't, but Rufus is coming up in conversations a lot. So I'm very interested to see what you're gonna say about it.

SPEAKER_00 6:56

Amazing. Yeah, I know it's way more popular in the US, um, but we'll dive more into the actual stats. So, Rufus, for anyone who's tried, can probably confirm. What's cool about it is not it's not just one way, it's not just a chat bot. So you can use text to prompt questions, get the answers, but you can also use voice, you can use photos, uh, and and it has price history, which can also help you make the right decisions, find the right deals. My favorite part of using Rufus is photos. I'm a very visual person. So a couple days ago, I just took a picture of the backpack because I want to buy the same, and it found it in seconds. So I didn't even have to look for the model, I didn't even have to describe what exactly I need in there. That's that's how useful it's becoming, right? And the best part about it is it has your purchase history. So it can actually give you the results and recommendations um that are more personal to how you tend to shop, what you generally prefer in in different niches. So that's Amazon's big advantage. And Andy Jesse, the CEO of Amazon, actually understands that. And a few weeks ago at World Economic Forum in Davos, he basically said, I'm very bullish on Gen E commerce. And he said that our advantage is no other LLM, no other tech company has all the buying history, right? Yeah, and rightfully so, he's bullish. Like Black Friday 2025, AI tools influenced 14.2 billion in global sales. We don't really think about it, but they already influence our decisions. Now, 3 billion of that only in the US, which is 805% year over year growth. And according to Amazon's report, about 38 to 40 percent of Amazon sessions during Black Friday used Rufus. Wow. Quite a lot, right? Rufus is Amazon's gateway to agenti commerce. And what is agenti commerce? In simple words, it's basically you don't search, but AI searches, compares, filters, and buys for you.

SPEAKER_01 9:02

Amazing. Okay. Does it because I mentioned at the beginning I I consider sometimes reviews when when you buy a product, would that be integrated as well?

SPEAKER_00 9:14

Yes, very much so, especially with Opus. So it relies a lot on the customer input, not only on the uh seller's input. Um, so yes, customer and uh customer reviews and customer QAs are definitely used in creating those responses, and we'll dive into actual examples a little later. So let me explain the whole customer journey with uh AI large language models. So it normally starts within 10, just like you said, hey, I have this problem. How do I solve it? Right, then LLM suggests some recommendations, and you will have an option to buy right there in that conversation. So that's the idea behind Agente Commerce. All of the sellers from like 10 years ago would probably say, hey, it's basically search, fine, and buy. That's what we call it in the Amazon's customer journey. So the idea is Agente Commerce will take care of the whole thing, it's all gonna be automated. You won't have to interact with the platform as much, you'll just have a conversation, right? But are we there yet? The simple answer is probably not. Just like you said, an average person mostly just uses answer engines to just understand what like what is the solution potentially, what direction should I go, and then still use what they're used to, right? So completing the purpose of the product within the conversation chatbot is currently the least adopted use case. Now, where we had it though is very interesting. So OpenAI is working on this ACP, which is a Gentil commerce protocol, which has in-chat checkout. They pulled back six months into it, uh so two weeks ago, they're restructuring it now, but doesn't mean they're stopping. Google developed a similar thing called Universal Commerce Protocol, UCP, with over 20 partners, and they also have live checkout inside AI mode and Gemini now. Okay. Alibaba in China is the only fully closed loop where they own the whole process. They own the AI, the marketplace, the payments, and the logistics. And that's probably why they're a little bit ahead in actually making this highly adopted in China, because they don't have to work with any other partners and integrate, which creates a whole uh bunch of issues on the technical side, right? Now, the most agenda commerce today is still mainly just discovery and assisted checkout. So human is still in the loop. Yes, for now.

SPEAKER_01 11:49

Well then things worried.

SPEAKER_00 11:51

People who sell on Shopify would have received an email like this. I got this uh when I was in Vegas for Prosper from Shopify. The email basically says, hey, your products will now be searchable on Chat GPT. They're not asking you for permission, they're just telling you this is gonna be the case, right? So we're we're definitely progressing there, whether we want it or not. The tech companies love this idea and they're all investing huge amounts of money and efforts into this. So we're basically at a similar inflection point to the early 2000s, where anyone who's watching this um will see the graph. But if you're listening, it shows the graph of uh Twitter, Orex, Instagram, uh, Facebook, TikTok growing over time, being adopted over time. And next to them, there is a graph for ChatGPT, and it's basically a vertical line. So ChatGPT is the fastest growing product we've ever seen in the history of humankind. If you compare it to mobile phones, ChatGPT almost reached a billion in about three years. For mobile phones, it took 20 years to reach a billion. Smartphones, a little faster, five to six years, but for mobile phones as the as the new way of talking and and still have getting life, 20 years, so quite a lot, right? But if we compare LLMs, all of the large language models to mobile phones, so ChatGPT, Gemini, Meta AI, Deep Sea, Perplexity, Gro, Copilot, Claude, Hagenface, Manus, in total, that's 2.3 billion users, which is one-third of the world's population.

SPEAKER_01 13:29

Okay.

SPEAKER_00 13:31

So it's definitely a new way to interact with the internet, right? And with with with the information. And it doesn't seem to be going back or stopping anytime soon. Um, for instance, AI assistants are already at 300 million daily average visits. A lot of users yet, but 300 million daily average visits. Chat GPT, 75% of that Gemini is on the rise. Okay. Okay. And this is fun for e-commerce sellers. So ChatGPT is now the number two product discovery platform. So it's basically 50% of the Amazon's traffic for specifically product discovery. Just like you said, Natalia, in the beginning, you could ask some advice, like hey, which gift should I buy? So Chat GPT is being used for this a lot. And if we think about this, if like a lot of people use other platforms too. So I have to Gemini Claude, Copilot, Perplexity, we're probably getting there to be comparable in terms of product traffic. So the shift is definitely happening. And when we when we look at the studies, for example, this one from Accenture, um, it clearly shows us that consumers already trust AI more than friends and family, social media, marketplaces. Um, so they're uh ask thousands of people what is your preferred source for purchase recommendations? And NAI came up as number two, uh, right below physical stores. Um and it's probably because it's more similar to how we tend to converse when we just go to a department store and any shop, right?

SPEAKER_01 15:05

It's that trust element. I know I know exactly what you mentioned. There's been such a huge shift that we ask ChatGPT anything or Gemini or Genai, and what it gives to us, we trust it. We trust it 100%. Um, whereas that you're showcasing, we always have a little bit of a doubt. Um, but that shift has just recently happened, and I I can't imagine how much is gonna change in the next couple of years.

SPEAKER_00 15:32

Yeah, 100%. And I think it's it's also because we're used to thinking, uh, hey, if I go to Amazon, reviews are probably a little bit manipulated. I don't know if I pause them, like nobody leaves five-star reviews. Why do I have 90% of them five-star reviews? All of that stuff, like people understand, and then they look at uh Chai GPT's answer and they're like, oh, it's all based on these sources, it's probably scientific, it's all research, right? Yeah. So we feel like it's it's more uh emotionless and less in uh influenced. But obviously, we'll talk about how we can influence it as well as as brands. So Amazon and OpenAI, very exciting part. Amazon recently invested 50 billion into open AI, a lot of money, right? Uh, what does it really mean? Amazon will tailor open AI models for use across Amazon's AI products and agents that serve customers directly. So it seems like Amazon is already using a lot of the open AI's technology already as a result of that transaction. That AI is reading your Amazon listings today. Okay. Doesn't seem like Amazon is sharing Amazon's data that that that they'll be surprised to see, but they're definitely using OpenAI's advantage. Now, for sellers who um are deep in it, understand how critical conversion is, you'll love this. So I'm showing a couple graphs from Sensor Tower. Uh so they they've done a study during Black Friday uh last year, 2025. Non-roofless conversion ended up being on average 11%. And it they compared to ruthless conversions, people using roofless before buying, and that conversion was on average 36%. So 3.5x higher conversion after interaction with the chatbot.

SPEAKER_01 17:15

That's incredible. Lucy, well, there's a lot of focus, I guess, on the US market right now. Um, you might not have the answer to this, but when do you think that shift is gonna come into the UK, Europe, other territories? I know US are always the first to do everything and then it shifts. But I'll be interested to see when we can get these kind of statistics in other marketplaces.

SPEAKER_00 17:40

100%. US is definitely their uh playground and all of the new things they test there. We'll dive into some other exciting things that are already live uh on Amazon US. Um, but it is it is live in um the UK. It is live in a few more markets uh in Europe. I believe it's Germany, Spain, uh probably some other markets. So they're like they confirm this is the way, and they're already slowly but surely implementing it everywhere else. Uh, but of course, US is the biggest market and will always have the biggest science-backed data uh and information from it. Um but realistically, it's it's gonna like Alibaba, China is is already ahead of us in the Western world. So I think it's inevitable for Europe to catch up. Here's something from NBJ. So the CEO of Amazon, who's apparently very bullish on Jeff ECommerce, um shared this report, Q4 2025 report that's public about Amazon's performance, which says over 300 million customers used Rufus in 2025. The monthly account users went up by 149% year over year, and the interactions went up by 210%. Okay, so way more than they expected, they projected a little less growth. Another interesting stat he's famous for, because he shared that in all possible media outlets. Customers who are using Rufus are 60% more likely to complete a purchase. So if he again says conversion is much better, right? Then Rufus generated about 12 billion in incremental annualized sales in 2025, and they project it to reach 10 billion part in the year. Okay, that tells you something. Here's another interesting step for the sellers. So if you sell in the pet care space or travel or entertainment, these are the top categories where, according to Bain and Company, uh people trust AI to make shopping decisions the most. So pet care is 46% of people generally talk to a chatbot to research and make the purchase decision. Travel or entertainment is 43%. So quite a lot of people actually already rely on AI. Okay. Another interesting part here, um, US consumers trust their familiar marketplaces much more than tech companies or AI-focused companies to automate their purchases. So think of subscribe and save. If you bought something before, uh, or if you just want to reorder again, like obviously people trust the marketplaces much more to do that again versus something new. So let's say uh Google's Universal Commerce Um protocol or OpenAI's uh identity commerce protocol, where they build the whole process into the chat conversation, they definitely have a little bit of a disadvantage there, right? Okay. Now here's another interesting stat for the businesses. So far from consumers adopting this, let's see how businesses actually look at uh generative engine optimization. So small businesses, surprisingly, are the least prepared and adopt GEO the least. So 80% of small businesses actually are doing some kind of GEO optimization, right? Medium-sized businesses, 21%, and enterprises are 16%. So that's NP digital research, and that's not why we focus on enterprise, uh, where there's definitely big interest, definitely big budgets go into these initiatives, um, but definitely medium businesses are still ahead.

unknown 21:16

Okay.

SPEAKER_01 21:16

Exactly. And is that just the cost? Is that why smaller businesses don't use it?

SPEAKER_00 21:22

Or I'm not too sure. I think it might be mindset as well. Uh, we tend to be creatures of habit, and a lot of people just don't accept that the changes here, I think. But realistically, the shift is we're transitioning from search engines to answer engines plus action engines. So action engines is something that's still kind of in the works, but answer engines are definitely deep into our lives already. Okay, good news. You're watching this, so now you know. There's no way back. And so so what Amazon sells would tell me like, what do I do? As an Amazon seller, what do I do with this, right? Um, so after After this, I can dive into the actual practice. But let me know if you have any big questions you wanted to kind of quickly chat about before I look at um the workflows with everyone.

SPEAKER_01 22:10

Yeah. I mean, what all our viewers are probably asking themselves is where do I start? Um, we understand the shift, we understand our customer patterns and the way that they work now, but how do you start implementing this into your business?

SPEAKER_00 22:26

100%. Yeah, that's that's a big question. And this is what I'd like to uh also structure in in into uh the three levers. So for Amazon sellers specifically, um the three levers that I currently know, and we'll dive into the science and exact action plan you can follow. Okay. So before I dive into it, quick story. I'm not the best cook, I cook once a year. Okay.

SPEAKER_01 22:51

Just one tie.

SPEAKER_00 22:52

Just one tie. That's on February 14th for the Valentine's Day. And I'm pretty sure my wife is happy I don't cook more often because I'm just not that good, self-awareness. So last time I asked this question, what is the best pasta for Pentic Home dinner? I asked it on Rufus. Yeah. Okay. This is what popped up. So I had uh Rufus suggests a few products, and I was also curious to see what the search results will show me. So I also searched for dry pasta. So if you ever's watching this, you'll see the screen. Uh, and the right side of the screen will show me the traditional search results on Amazon. Left side will show me Rupert's results, and I kept scrolling through it. Left side had other interesting products, the right side had usual search results, and not even top-rated or best sellers showed up in the Rupertus recommendations. I was like, huh, interesting. Okay. I kept scrolling, no match whatsoever. And that told me that left and right, so Rupertus versus traditional search are completely different algorithms. And the big question I have to you as forward-looking entrepreneurs which part do you think has bigger potential to get investment and get developed and get adopted in the daily lives of the shoppers? My personal bet is the left side, the genetic side, the Rufus AI in this case. And if that's the case, the big question is how do I show up in Rufus recommendations more often? Right? That's what we'll um outline in this process. So the three levers to boost AI visibility. Lever number one is Cosmo. To some people that might sound familiar, Cosmo is the knowledge layer. Okay, so that's the new algorithm Amazon is relying on, not solely on it, but heavily relying on Cosmo. Okay, lever number two is Rufus, so the conversation layer, and lever number three is citations, the trust layer. Amazon is also using external sources. Now let's dive into lever number one. Lever number one, actually, before we do that, quick history on how Amazon product display page optimization evolved. Uh, for those of you who sold between 2010 and 2016, you can probably relate and confirm. I call this era listings 1.0, so keyword games. It was all about keyword stuffing, review and rank manipulations, right? Then next era, listings 2.0, 2017 to about 2024, that's about content and brand. So the better the content, the better the reviews, the higher your click-through rate, and the higher your conversion rate, and the algorithm is built to reward whatever sells best, right? Now, 2025 onwards, in my opinion, it's a new era. It's called I I call it listings 3.0, so AI visibility era. So that's cosmore understanding, ruthless answering questions, which also index conversion, and external citations validating. And this is what we'll focus on today, this specific new wave. And quick note both of the last era, so listings 2.0, listings 2.0 coexist. So you don't need to ignore whatever you learned to optimize for highest conversion and keyword ranking. It still works, but please be mindful, this is not the only way, and it's not the way that gets developed the fastest. Okay.

SPEAKER_01 26:23

Yeah, that's a great point because I think some viewers might be thinking, okay, with this shift, what about all the efforts I put in the past? But actually they were worth it and they will have an effect on what's coming.

SPEAKER_00 26:36

Yeah, at least for now, uh, based on what we've studied. Don't ignore what you've been doing. Um, most of it probably is still critically important, right? Here's another validation for where Amazon is pushing this game. So some of you may have heard who are into PPC and love optimizing uh the ads side of things, the paid ads. Amazon launched sponsored prompts, and on March 10th, I got a message from uh Amazon's own product marketing manager, uh, who I actually met uh in Milan on uh Amazon ads creators retreat. And she's like, Oh, I know you're into this, you post about this stuff. This is coming out soon, uh, but I'll I'll message you before it becomes um email, newsletter famous, right? And it's basically the the announcement that from March 25th, it's no longer gonna be beta, it's gonna be out for everybody, and people will be paying for these ads. So, what are these ads? I was seeing the screenshots, you'll see um something that looks like Rufus suggested questions on a listing, but one of them is sponsored, right? So that sponsored mark is there, and when you click on it, there's a conversation, and you as a seller will be paying for this now. That conversation is built on the information you provide as a brand. Okay. Uh, and I look into some reports from the customers we work with, and I'm I'm sharing the screen as well with a report that will look very familiar to people who love BPC ads. It has the click-through rate, the cost per click, the spend, 14 days total sales, all of that data. I already see that there are sales happening from people clicking on these sponsored prompts. Again, for now they've been on data, so for free, amazing. But from March 25th, we're gonna be paying. Um, that's why on the screen chart you see there's zeros for now, so we don't have any spend, but it's working. And Amazon's staff told us in person basically it's it's working better than we expected. We're gonna release this, we're gonna expand this. Uh, these ads will be on competitor listings, and so on and so forth, right? Uh, where you find the support, you go to ads console, campaign, ad group, ads, and you'll find prompt tab there. So that's fairly new. What we currently know, as I said, March 25th, 2026, it becomes paid. Amazon writes the prompt and the response. That's something new for PPC optimization gurus. You write nothing, right? So it's all based on the listing content. So it's literally your ad creative now. So weak listings will lead to weak prompts and you'll be paying for bad answers, right? So, number one thing I I think you need to do now. Um, you most likely have this on already. So audit and pause individual prompts in the ads console today. So before you start paying, make sure you're only paying for the ones that actually get the clicks.

SPEAKER_01 29:29

And as you change the prompts, well, as you change the listings, will the prompts and the responses change as well? So is that going to be adaptive post-March? Sellers uh can still change that to optimize it, basically.

SPEAKER_00 29:43

Alrighty. So yeah, you're exactly right. We we can optimize the listings uh for um rule proofs, basically, uh seeing all the questions for the copy with images, and that's the information, including uh brand stores, including your keywords in sponsored uh product campaigns and sponsored brand campaigns. All of that is being used um by the sponsored prompts to generate these prompt responses. Yeah, so that's that's the indicate. You need to optimize a little differently now, uh so not only for keywords. So explain sponsored prompts in a graphical way here again. For anyone listening, I'll explain it. Um, level one is Cosmo, as I said, level two is Rufus. So we optimize for it. So then sponsor prompts, which is the paid amplifier, uh, will say everything louder, right? So Cosmo is what Rufus knows, then Rufus is what the chatbot says, and sponsor prompts will just uh enhance the sound and the loudness of that message. So let's dive into the actual way we optimize. Lever number one, Cosmo or the knowledge layer. So a lot of sellers kind of heard that name by now. Amazon Cosmo, right? So what is it really? Um I'll I'll break down the optimization into these three parts. Number one, we need to make sure we're in the in the right browse node, okay? Browse node and also item type integrity. Uh, so that's the number one thing that describes a product. We have to make sure it's correct. You can't really play these games anymore where you just put your product into the subcategory where you have the best chances of being the best seller. It has to be the closest to what the product actually is. Number two, backend taxonomy. Uh, number three, there are 15 relation types, uh, and we need to optimize for noun phrases. I'll explain everything shortly. So, number one, browse note and item type integrity. So you just have to check this your ASIN is in the correct browse node. Sometimes I see sellers just place them in in not really relevant subcategories. So you have to fix that. It's not gonna help you anymore. You have to put it in the right um browse node. Then there's item talk.

SPEAKER_01 31:51

How would this negatively impact you? Let's say that someone has accidentally not done this and they haven't put it into the right category. Are we gonna see a huge negative effect by doing that?

SPEAKER_00 32:03

Absolutely. Yeah, and thanks for asking that. When uh the product, um, let me think of a good example. Let's say I'm saying uh chilonic acid, beauty product, right? And I placed it in um creams. Now, anyone asking for the best childrenic acid, whatever criteria they have, let's say vegan chiluronic acid, and let's say I'm selling vegan chiluronic acid, I'm optimizing all the keywords, but if I'm in the in the eye cream category, for example, somehow, whether it's by choice or just happen because things happen on Amazon, there's no chance Rufus will recommend you because you're not in the right category. That's as simple as that. Yeah, so thank you. Yeah, huge thing. That's why I'm listed as number one, because you do everything else, it it won't matter, right? Um now another like it's probably the number one thing in the backend attributes uh that you'll see. Uh it's item type keyword. Another important thing that you just need to explain the product properly um without over exaggerating what it is, and also without understating what it does. So very important to also have it right uh because it's one of the basic things, answering one of the main questions Cosmo is collecting data on, which is what is this? Right? That's not the critical thing. Now, the way number two is the backend taxonomy. So expanding on the rest of the back end, stuff that people don't even see on the listings on Amazon. Um, you want to optimize back-end attributes as your priority. Uh, and I built a prompt, so anyone who's using Claude can I'll share uh QR code at the end, can download those skills. Uh so skill is like a file that has the prompt built in it, upload it to your Claude, and use a very, very simple prompt which basically says use Amazon BA, which is the name of leaks skill, Amazon BA skill, to optimize the attached backend attributes for Cosmo. So all I do in this process, um, well, obviously Azoma has a different suite of tools, but this is what is this built on a side when experimenting. All I do is I print out as a PDF the product details page of the backend. I'm not as technically advanced as some other people that just download the flat file, so that's also an option. But I just have a PDF attachment, right? I upload it to my quad and I use this skill in that very simple one-line prompt to create something like this. So this example where I was watching will show you a table that that is my output of that prompt, and it shows me all the attributes and the status, whether it's weak, missing, or it's perfectly fine. What is the current value in there? Why it matters based on Cosmo uh science and the recommendation of what needs to be changed, and then the new suggested input, which is an example of optimized value. So that's just a simple thing you can do. Uh, I already have this ready for you, you can just use my process, and it's gonna help you optimize the answers. What I see a lot with customers, including huge CPG brands that we all have at home, their listings backend has lots of missing values. And when information is missing, or if it's not precise, or when you say in the audience adult, and that's all you say, but it is actually specifically for females 35 to 55, you're missing a lot in in terms of being understood by the new Amazon's algorithm. Does that make sense?

SPEAKER_01 35:40

Yep, definitely.

SPEAKER_00 35:42

Amazing. Let me dive into something else for Cosmo. So the way you optimize for it and the way this skill knows what's missing, what's weak, is all based on Amazon's public science. So there are 15 relation types that Cosmo paper, so official white paper from Amazon, has outlined and explained. There's a QR code for anyone who's watching, you can download that PDF that basically is Amazon's um white paper explaining the science, which is amazing. Nobody else really shares that. Walmart doesn't share their algorithm with us. Amazon's got enough to give us uh that background. I'll dive into one quick example that's uh interesting in there. So they they they show this case study in there to explain the algorithm. They said, hey, here's pregnant women looking for shoes. She ends up buying slip-resistant shoes. So now Cosmo understands that pregnant women are looking for slip-resistant shoe. So the listing of the shoe doesn't talk about the shoe being for pregnant women, but now there's this connection, right? So pregnant requires slip resistant. That's that's kind of how it works, which makes it kind of interesting. So it's not really keyword-based anymore. Now, something else they listed is the actual relation types. So there are 15 in total. Doesn't mean all of them apply to your product, but these are the ways to describe the product, right? What is the function, what is the event this product this product is for, what audience it's for, um, and what body part do you apply this to? Uh, what is this complementary uh to all of these kind of things that help explain all possible sides of the product, right? So make sure that your copy also answers the key questions that apply to your product from this list. Okay. Pretty interesting, right?

SPEAKER_01 37:31

Yeah, so if the the when the pregnant woman searched for those shoes and then the slippers showed up, do I understand correctly that then the next time that the question is asked, it's based off of previous prompts, and that's how it just keeps adapting and evolving?

SPEAKER_00 37:50

Yeah, so part of it is Amazon's system now learned that regardless of these products mentioning pregnant women is the audience, it is the audience now, right? And it's gonna be shown as a priority. That's how they personalize it too. So my results for whatever I'm searching will probably be a little different than than yours, Natalia. Because my shopping history shows different behavior, right?

SPEAKER_01 38:13

Yeah, it's I mean it's all tailored at the end of the day.

SPEAKER_00 38:16

100%. And that's that's Amazon's advantage, and that's probably why uh I'm so bullish, at least on uh Amazon's agente commerce play. Now, so far from what I explained, do you understand the difference between Cosmo and Rufus?

SPEAKER_01 38:29

Yeah, I mean the theory makes sense. I guess what everyone's question is is the practice. Um I know when you explain it, it makes sense. It's really nice and clear. It's just putting it into practice. I guess that's maybe where people will struggle or will maybe need some help.

SPEAKER_00 38:46

Yes, and that's uh that's why I uploaded these files that I'll share with you at the end to where it kind of simplifies your your work a little bit. You don't do this manually, you don't have to in 2026, um, where the the the the chatbot will just tell you, hey, fix this, you know, based on the all the science that I know. Now, quickly to outline differences, Cosmo is the common sense algorithm. So it answers the questions like what is this product, who it's for, why people buy it, right? Pregnant women, slip uh slip resistant shoes. And Rufus is the shopping assistant, so just a robot helping you, right? Which product answers your question? What is the best for you? Right now that I can tap into Cosmo, not only Cosmo, but Cosmo is the big part it relies on. Uh, what is the best for what you're asking, right? So if we outline the uh the match, Cosmo and Rufus, uh the the slide looks odd, but that's fine. So I'll go through it. Cosmo talks about product meaning, text attribute extraction, use case mapping, right? So use case, pregnant woman, slipper uh slippers issues, common sense tagging, and then Rufus on the other side is all about intent matching, reviews and QA synthesis, visual label tagging, so that's where we optimize the images, and then answering and ranking. And together they meet for the product and intent alignment. Okay, makes sense, right? Swing. Now let's dive into that second huge lever we can optimize for, which is Rufus, the conversation layer. So Rufus is something we say left and right uh all the time. Uh like LinkedIn on Amazon's topics is half of it, it's basically talking about Rufus. How do we actually optimize? So I suggest three steps. Number one, we audit, so we collect niche questions and we test the answers on our listings, right? We need to collect the data. Number two, we enhance our copy, our images, our A. Answer these top questions. And number three, we seed proactively populate questions and answers. There's another section for that. Uh, so Amazon can also rely on um customer outputs and all possible ways we optimize. I'll show you some examples to explain how to seed QA. Now, number one, we audit the niche. How do we audit? We collect the actual questions that Trufus is suggesting. So anyone who's been using Amazon in the last year or so would know that under the image there's a list of questions that just pop up as suggestions, suggestions for how to start the conversation with Rufus. So if you click on any of them, the answer will come in, you can ask more questions, and that's how you start talking to Trufus. Another part is right above the reviews, um, it starts with like looking for specific info. So that used to be QA section where people just ask and answer questions. Now this is also occupied by Rufus, and it sometimes has even more questions, so other questions. All of that is your like data playground. And what I suggest doing, and what Azoma has automated as a software, you go through all of the listings that matter in this space and you see what questions are being suggested. Because these are the frequently asked questions on these listings, and your listing will very likely have slightly different questions than your best-selling competitors. That's why you need data on the whole niche. And whoever sees this on a video will see a screenshot on the left of uh one of the tables. It's just an export from Azoma. And I think over like a few weeks we collected data on how often all of these questions show up. And in this example, there's one that says, can it be used daily? And that was asked, uh asked on the listings we collected data on 447 times as the number one most frequently asked question. So now we know that daily use is a critical thing to answer, right? And we have hundreds of these questions, and our goal is to identify which ones matter the most, which ones are the most relevant to our product, which ones do we need to properly answer in the copy and in the images. Does that make sense?

SPEAKER_01 42:51

Yeah, perfect. And then would you recommend that you do the same thing with your competitors and see what questions are being asked there and then tailor it that way?

SPEAKER_00 43:01

Yeah, so when I'm looking in into the niche, uh, I'm looking at my listing and hundreds of others that I know are in my space. And sometimes they'll be different enough for me to ignore these questions, but very often we'll have a lot of common questions that I'm like, oh wait, this didn't show up on my listing. Uh, like maybe is this shampoo good for curly hair? It didn't show up on my listing. But it's definitely important to answer it. I I get it now, right? And that's the data Amazon gives you, so we just need to collect it and organize it.

SPEAKER_01 43:31

Amazing. Okay, so once you are the tips, yeah.

SPEAKER_00 43:36

And another quick tip before I dive into the action step with this data. So, something you asked earlier on, Natalia, is does it look into reviews, right? So here's a quick check. If you ask a question and the answer comes in with something like most customer support, most customers say many customers experience this. That means the answer that Rufus just combined came from customer inputs. So that came from reviews, and that came from the QA section. Okay. So when you see that, there's definitely like it's a big, big flag uh saying, hey, there's something you can optimize. Uh, your inputs as a seller are not used. So you can optimize your listing better in answering this question. Uh, in most cases, sometimes the question just leads to stuff you can compliantly state on your on your listing, and that's why you would use some other ways to optimize that I'll explain later. Now, um, yeah, I think I think that's that's it for for this one. How do we actually optimize? So here's the enhanced part. So copy images and A is what we can use to answer the top questions. So starting with copy, we want to answer these key Rufus questions. I also created this file to make it easier for everybody. Again, this is not a Zomas uh file because we have a software built for this, but to make this easier for you if you didn't just use, you can download my file, do the same. So you would use this skill to enhance the bullet in the description. The skill basically says, hey, based on the uploaded list of questions, so that export I showed before, uh, for this specific product, address the top 15 to 20 highest appearance, most relevant questions. And it's basically going to create an alternative copy for the for your title bullet description and still review it manually to make sense, it's not hallucinating, and that's that's how we can quickly make the change. Example of the output would be a table like this. Uh, so we show you the current inputs, the suggested ones, and explain science on the right for both Cosmo and Rufus. That that prompt also covers Cosmo.

unknown 45:48

Okay.

SPEAKER_01 45:50

And you would recommend doing like a combination of the both. Obviously, you have your existing one, you have the prompt, and then just review it and slightly change it so that it aligns with everything.

SPEAKER_00 46:01

Exactly. The way it's built is to make minimal changes. We don't want to mess up the SEO, so it's normally just going to add a little bit. And maybe change the sentence a little bit. So by adding curly hair, now we're addressing the curly hair question, right? Stuff like that. Um, and the the the other part is images. Images is huge, optimizing images is not easy at all. As a co-founder of productinion where we A-B test images, um, former co-founder, Clarity. I know it's a lot of work. So the goal here, there are two things. One, we want to frontload key information by reordering uh the images. Uh, even for humans, it's important to speak about the most important things first. And there's another skill that I'm sharing here, another quick prompt you can use to basically look at the questions people ask on Rufus, look at the images and understand which ones matter the most. Okay, another quick thing you can do, so definitely use it. And the outputs would be something like this: the current order, the suggested order, with a number of words, like the old position in the brackets, uh, so you know which exact image moved. Um, and these images are named by image types to help you understand if this makes sense. Now, an image is definitely worth a thousand words. And especially an image like this, what I'm showing is uh an infographic uh with a remote controller that explains all of the technical data, all of the major questions people ask uh about the technical specs of this product. And when I was looking for my remote controller, this image was shown by Rufus a lot. I would ask question after question, and I'm but still see this image. I'm like, that's pretty cool, let me just look at it. And images definitely matter more than just text. So people seeing images are more likely to convert. And this is basically how it works: there's multimodal understanding built into Rufus. Rufus taps into another algorithm that Amazon is using, it's called Amazon recognition. It's built to recognize what's happening on the image. Okay, so it recognizes the text, the objects, the person, the use cases, and the context of the uh image to match shopper intent and generate the best answers. So we see the face, we see what emotion is there, what kind of person it is, what age group falls into. So make sure your lifestyle images match your Amazon reports for who actually ends up buying the product. So you don't show elderly people if it's for kids or whatever, right? So it's a good way to check. Here's an example of how it works in practice. I asked this question on Rufus: is this ship good for curly hair? Which is one of the big questions. And the image itself doesn't speak about curly hair necessarily, but it's recognized that the person on the image has curly hair. So the text answer still came in answering my question in a proper way. Image was just more like, hey, I know if I show the image, there's high likelihood Andrew is gonna convert. It looks like him, potentially, if he is asking about curly hair, so he's gonna relate, right? So that's that's confirmation that uh Amazon uses uh that algorithm for Amazon recognition, uh, image recognition as well. Now, quick thing to do this if you don't want to create a WS account and um upload images that way, which is also an option, when you add a product, so you don't necessarily have to end this process, but if you click catalog add product and you click upload images, and then you click generate content, it'll generate a title only based on the image. So Amazon will read what's on the image and will suggest the title, right? Um, which tells us how well the image communicates what's on the image, right? So in this case it was still the same remote controller, but Amazon thought it's a portable power bank. So knowing this, I'd probably look for ways to potentially add a package in the background that actually says what this is, so it's easier for Amazon to interpret this correctly, you know. Quick exercise. Yeah. Now, another last thing for images. How do we maximize um what what is called OCR, so optical character recognition? So the lowest hanging fruit uh for image optimization is adding some text to clarify what's happening on the image, to answer some questions, uh, because most of the uh most most of the source still have just pure image, no text whatsoever. And it's very easy to edit, right? So images are subjective, what's beautiful, what's not is subjective, but adding some text can help you really get understood by Rufus uh better. So there's another prompt you can use in the same chat bot because it already has all the Rufus questions, and it basically says for each image produce a table uh with like um the current image, the suggested text overlay, um, Rufus questions we answer with these uh additions. Uh so useful thing you can play with this as well. Uh you can probably improve that prompt too, but that's uh something I prepared for you to make it easier. Example of the outputs um would be images on the left, middle column suggesting what text to add on the image, and right column listing the questions we're actually answering from the Rufus data. My favorite are something like the last example here, single feature, which is just an image of a shampoo on white background, says absolutely nothing, complete loss of real estate. So we can easily add some call-outs. And in this example, we answer five different questions. And again, when people see images in the roofless answers, they're more likely to convert. So definitely take advantage of this.

SPEAKER_01 51:33

Perfect. Yeah, it's going back to that thing that you said about trust. Um, as soon as you see an image related to what you said, that trust builds even more, and then you're more likely to buy the product.

SPEAKER_00 51:44

Exactly. Yeah, you're spot on. Now, last thing we can optimize is A. Okay. So A plus has still allows us to do alt text, at least in the US. I've seen in some other marketplaces uh Amazon is already automatically doing the alt text, which is basically text explanation of what's happening with the image. But in the US, as of today, you can still do this. So definitely uh go to your alt uh A plus, click on edit with images, um, and provide 100 characters long explanation of what's on the image. I normally do this in the same chat conversation. So then Claude has backgrounds for all the questions that matter, uh, and the the 100 character lengths fits some of the answers in there where it's appropriate. So I wouldn't lie about what's on the image, but I would make sure it's addressed in it in a proper way. Now, another thing that I recommend everyone do at this point is adding, or if you already have it, editing your A QA section. So A plus has the section that any seller can add that basically lists common questions and answers. Um, and we want to also rely on the data we collected for Rufus. Uh, and I have another prompt here that you can get from the slides to suggest how do I in the brand voice ask the most common questions and answer as well in a proper way following my own brand guidelines. Yeah, this is what it looks like. Very simple format. Uh, it's at the bottom of A normally. And now the third part is something that sellers have done for a while. Um, it used to be customer QA, now it's taken by Rufus as well. It's right above the review section. Um, and we can see the answers there. So there are multiple ways of doing this, and again, I encourage people not to um try and skew the algorithm or do any kind of black hat stuff. But the absolute minimum you should do is go through these. If you click on any of the suggested questions, it'll give you the answer generated by Rufus. And it's mainly based on reviews and the QA, so questions and answers done by customers previously. And if you click on this that line item at the bottom, show related customer reviews in QA, Amazon will actually show you what's been used to generate that answer. So now you know what impacts that answer, right? And if you click on see all of these answers, so see all two answers for that number one question. In this example, uh, you'll end up on this page where you can answer it, and you can answer it better. And as a brand, you should be answering all of these questions anyway. So definitely answer them. It gives you another layer to optimize for. Um, and as we know, Rufus relies on both seller inputs and also customer inputs. But this is how you can, as a seller, still impact the customer input section, right? And when you do answer that, at least uh the format I'm using is um I'm trying to implement at least one Cosmo relation type into it to maximize my chances of Rufus picking it up because now it has more L information. Um, and we believe currently, based on the testing we've done, that these answers will be prioritized. As soon as you have new questions, uh, but even like normally they don't show up here because people just talk to Rufus at this point. Uh I I I encourage everyone to just go through all of the questions that they are there. So click through all of the Rufus questions, look at your competitor listing, see what questions Rufus suggests, paste them there too, and see what shows up in this section and see if you can answer that better as a brand. Okay. So yeah, to summarize, we can add questions to listings, we can submit brand answers, and Amazon customers will also answer that. So if there is a new question at some point, it notifies Amazon customers and they come in and answer too. So just keep that in mind. But our main goal with this is we want to expand the data to REM. So now I'm ready. If you are ready, I'm ready for the last lever. Citations. Let's do this. Okay. So citations is the trust layer. Okay. This is how Amazon validates the recommendations and minimizes the manipulations sellers tend to do. So the trust layer, um, we're gonna break it down into three major steps. Two, we audit, so we collect the data to identify which external sites Urpus is already citing. Okay. Number two, we enhance, so we build our own websites, PDP pages with full specs, and we also publish blogs that answer top of those questions. Okay. And number three, yeah, we seed, so we approach affiliate media with affiliate commission offers, and we send them drafts to make it super easy to review our product and publish about us. Because outside of Amazon's media matters a lot, as our tests have shown. So, can we actually earn Rufus recommendations to media? Simple answer is yes. How do you test it? Um, the screenshot here that was watching will show you what is uh an example. The question I asked Rufus: what is the best link design remote control for presentations with sources? So when you add that at the end, Rufus will actually show you the sources. Okay, you can click on the sources, it's gonna expand, it'll show you the actual links, and you can open these pages of the media that Rufus relied on in helping generate the recommendations in this conversation.

SPEAKER_01 57:28

Okay, but this will only show up if you put that prompt.

SPEAKER_00 57:33

Exactly. If you don't, you're not asking to show the sources, but Trufus is still using external sources to validate. Yes. So now you know which articles, which media from the outside of Amazon helps Rufus give you the recommendations. Side note, you may still see, uh or probably will now start seeing because I've been seeing this now, top direct products. So meaning products that are not on Amazon, but Amazon is still recommending them. Very interesting, right? So if you sell a lot off Amazon, drop me a message, I'll give you the email for Shop Direct program, and you can ask, hey, what do I need to do to be recommended by um by Rufus as well? So that tells us that Amazon is not limiting limiting itself to Amazon's ecosystem, it's it's collaborating with outside of Amazon's sources too. Okay, but back to the sources. I outlined here, so they're practically the 10 signals that make a page Rufus citable. So this is based on um the research we've done with the ZOMAS of collecting lots of data on what actually is being used by Rufus' sources. So number one, the URL slot has to contain the exact query. For example, best shampoo for curly hair. It has to be in the URL. Same thing in title, it mirrors the query. So if people search for best shampoo for curly hair, you want that in the title too. Number three, year has to be included in the title. So for example, if it's the latest and greatest of 2026 top picks, it's recent, you specify the year, Rufus will prioritize that article. Date has to be recent. So last updated date, if it's been updated, has to be also visible on the page. Then site has to be clear, niche, or relevant topic focus. So if we're speaking about beauty products, beauty magazine, specialist blog, something very, very relevant. Site has to be indexed or cited elsewhere on the website it's not brand new. And seems like Rufus is kind of different from ChatGPT, Gemini's, all of these LLMs. It prioritizes listicles or rank comparison format. So you know these articles that say top 10 best shampoos for curly hair, topics for vegan shampoos, whatever it may be. These are the articles Rufus loves. And each product has to get a distinct blurb, so not just the name and link, but explanation. Hey, this is why we recommend it, right? And product descriptions with use cases which connect us to Cosmo and all the attributes on the back end attributes, for example, ingredients, the benefits. So without that, these articles tend not to be picked up, right? And number 10, and something I'm very excited about is the products link are usually Amazon PDPs. So these are affiliate links, they have affiliate tag, and they more often than not, lately. I think all of them I've seen that have Amazon Associate disclosure. I think that's in your requirement. So these media articles will have that hey, we're Amazon Associate, that's why we're doing this, right? So what do we do with this? Now we have the checklist, we know what works. Um, a little bit of stats to understand the importance. The brand website. So my own brand can have a website, they tend to still be so uh used as citations, but on average 10%, 10.9%. Earn media about 16% from the data we collected. Uh and affiliate reviews, 72.7%. So for us to optimize the brand website, so at least something we have full control over, we want to ensure we have the PDPs, right? The actual product pages with all the specs that are important. Now we know what matters for Cosmo and what questions people ask the Rufus. And we want to also regularly publish uh blogs on how to use it or best for to help us enhance the information uh around our product too. But we'll focus next on the biggest part. So the affiliate review sites, which is 72% of the weight on the algorithm. Okay, so guess what helps with placements and affiliate review sites? Any ideas at all? Uh it's kind of part of it, but I'm specifically uh speaking about creator connections. So now they're another Amazon program. And you know, Amazon loves using their own affiliate uh their own um ecosystem for everything, they want to be independent, so they wouldn't use Slack for their internal channels, for example. They have something else they built in parallel. So, same here. They want to use everything that they already built. Creative connections is your shortcut, so leverage it. And the question I have to the brand owners who are listening do you already use creative connections for your brand? I know when I spoke about this like a year ago, most people were like, What is this? Now people do, right?

SPEAKER_01 1:02:21

Okay, so it's slowly increasing. Um, a couple of weeks ago when you were a prosper, how many people use this currently?

SPEAKER_00 1:02:29

I'd say about 20% raise their hands. Yes. And it's uh only still available to Amazon US sellers, uh, so will not be available in Europe as of right now, but I think they're rolling it out soon. So, what is this? I'll give you some um seller uh seller central screenshots here. So creative connections brand one example. So this is the brand I used to run. In the highlight here, you'll see 5.1 million as a total revenue from Amazon Creative Connections in year one. So that's extra revenue we brought only through Creative Connections, and 27.6% was the spend on the commissions. So if you compare that to PPC, A costs, like average cost of sale, these are good numbers. So if especially if you bring new to brand customers, if it's say 30%, that's that's pretty good. You're doing well, right? And in this case, brands annual revenue on Amazon versus what was brought by Creative Connections 88% was was normal sales, 12% came through creative connections. So it can become quite a lucrative sales channel, especially if you're in specific categories where influencers and media tends to talk about it, right? So maybe if you sell something very, very niche or something very, very technical, won't work as well. But if you're in beauty, supplements, headcare, all of these spaces where opinions matter, definitely take advantage of it. Now, another example, brand number two, first year 2.4 million in revenue from creator connections. In this case, span is a little higher, 32% was paid in commissions. But in their case, 36% of their revenue came through Creative Connections. So quite a big channel. And they they have competitors that just do nothing, right? So what do these brands have in common? Any ideas?

SPEAKER_01 1:04:22

The type of customers, um, I guess that's who influences, so that too, yes.

SPEAKER_00 1:04:29

And uh they happen to be both in booty space, but there are other spaces where creator connections work too. But what I was trying to uh highlight here is the majority of the creative connections content is article-based, and these uh brands are showing up in Rufus as a result of this very, very well. So even though we think about influencers as hey, it's a TikToker, she's recording the before and after, she's hyping us up. Most of the sales in career connections, in in these two examples, at least, from my own experience running them, came through articles. So media actually making money of the commissions. And I have a quick framework anyone can follow to maximize the chances of leveraging career connections for Rupert's recommendations. And all you have to do is number one, when you reach out to this media that you see is showing up in sources by Rufus, send them a draft copy. So make it super effortless to place you as number one top pick for that problem. Okay, and I shared previously that 10 signals that make a page Rufus citable, reverse engineer it, that's the checklist they need to follow. And most of them already do, but it's a good thing to check and you can help them out too. Number two, offer them decent affiliate commission. And you can do that for their connections because classic Amazon commission is one, two, three percent, right? They cut it substantially in the last two years. So the only way you can offer them much more lucrative commission is creative connections. So do that. Tell them, hey, I have this Creative Connections campaign, offers you 25% commission, 30% commission. Then they're much more interested to place you as number one, number two, number three, right? Now number three step, and that's something most sellers don't really do, um, is you you kind of want to build relationships. So check in quarterly, make sure your articles are updated, because as we spoke about RuPaulus, recency matters, right? So you want to make sure it's it's updated all the time, and when it's updated, the date is specified and your product is always ranked first. So if you're always on top of it, nobody else can do this and place themselves above you, right? So do that as your checklist for how to optimize for external citations. And our goal here is to own what others say about your brand and this way influence Rupus, right? Now, this is more to the audience listening. Who is excited to implement some of the actions? If you're quietly saying yes in your mind, you're saying yes, Natalia.

SPEAKER_01 1:06:57

Yeah, yeah. I mean, I'm excited for them to try it out, see what comes out, see what changes they can do.

SPEAKER_00 1:07:04

Yes, I'm definitely excited because I see a great result from our work um at Azoma. But here's a quick summary, just to outline everything we discussed. The three levers to boost the i visibility. Lever one is Cosmos, the knowledge layer. We want to make sure we're in the right browse node, make sure we answer the item type keyword correctly, make sure we optimize backend attributes properly and follow the 15 relation types that Cosmos uh white paper has shown us. Number level number two is Rupus, so the conversation layer, we want to collect the questions that show up on the um competitor listings and test these answers in our listing, and then optimize our copy, our images, our A to answer these questions better. And then we want to populate Q ⁇ A section with complete answers too. Level three is citations, so the trust layer. We want to identify what truthless actually cites in answering the questions, build our website's PDP page and write blogs, and also approach affiliate media with creative connections and Bashio Software. That's the plan, that's how you do it. I know it's a lot of work, um, but it does work. It helps you get ranked on Ruthless. In simple words, if your time and energy is limited, focus on lever number one, versus the Cosmo level. That's what Rufus knows about your product. Because without that knowledge layer, levers two and three just have nothing to build on. And to make it easier for you, this QR code will get you my skills that you can upload to Claude, these files that already have the prompts built into it. They're not perfect. Uh up will probably optimize them at some point. Azoma has their own suite of tools that does all this work, but you can you can probably play with this. And a lot of people do use Claude now. So hopefully this helps you um leverage what I've discovered so far.

SPEAKER_01 1:08:51

Amazing. Well, thank you so much. I think there was so much out of there that our viewers can actually pick up, take with them, make changes. As you said, this this topic is ever evolving. So I'm sure we'll hear more about it in a couple of months, a couple of years. Uh, but that's what's so exciting about it. So it's it's back to that trust, back to that element of using AI. And then hopefully everyone watching goes through all the steps. And thank you so much for all the um help that you give them as well. I think all those prompts will really help them.

SPEAKER_00 1:09:24

Absolutely. And again, I just played a lot with this, uh, and these prompts do help, but I'm sure some people will reach out to me like, hey Andre, you messed up here. You can do this better. That's how we all level up and learn from each other, right?

SPEAKER_01 1:09:37

Amazing, perfect. Well, before we finish off, I did have a question for you. In the end, what pasta did you bake?

SPEAKER_00 1:09:46

I ended up baking something else. Can't remember what it was. Probably wasn't as good as pasta would be, but uh, I did use uh uh Rufus for that other uh food selection too and validated with Chat GPT to be honest with you. Uh and basically the whole recipe was built by AI.

SPEAKER_01 1:10:04

Amazing. Well, at least that one meal a year worked out well.

SPEAKER_00 1:10:09

Yeah, I think that this time I didn't mess it up.

SPEAKER_01 1:10:13

Good. Well, thank you, Andrew, so much for joining us today. This was incredible. There's so much information that everyone can take from this. Um, so we really appreciate it, and thank you again for coming.

SPEAKER_00 1:10:24

Thank you so much for having me. Thank you.