Andri Sadlak is a serial entrepreneur and Founding Head of Product & Strategy at Azoma, building AI-first commerce software for the age of agentic shopping. He’s been an Amazon seller since 2017, exited his own brand, and co-founded ProductPinion. Today he’s part of the team behind Amazon growth and AI visibility strategies for 8 and 9-figure brands like Mars, L’Oréal, and HP. He’s one of the sharpest voices on Agentic Commerce, taking the world stage to break down how brands need to show up in the age of AI shopping assistants like Amazon Rufus, Walmart Sparky, and LLM search.

Highlight Bullets

> Here’s a glimpse of what you would learn….

In this episode of the Ecomm Breakthrough Podcast, host Josh Hadley interviews AI-driven e-commerce expert Andri Sadlak. They explore how AI is revolutionizing online shopping, focusing on Amazon’s shift from traditional search to AI-powered answer and action engines like Rufus. Andri shares actionable strategies for brands to optimize product listings for AI, discusses the importance of image and Q&A optimization, and highlights the growing role of external citations. The episode offers practical tips for sellers to thrive in the new era of agent commerce, emphasizing the urgency of adapting to AI-driven changes in e-commerce.

Here are the 3 action items that Josh identified from this episode:

Tools and Technologies

Studies and Reports

Skills and Features

Websites

Books

Key Concepts

This episode is brought to you by eComm Breakthrough Consulting where I help seven-figure e-commerce owners grow to eight figures.

I started my business in 2015 and grew it to an eight-figure brand in seven years.

I made mistakes along the way that made the path to eight figures longer. At times I doubted whether our business could even survive and become a real brand. I wish I would have had a guide to help me grow faster and avoid the stumbling blocks.

If you’ve hit a plateau and want to know the next steps to take your business to the next level, then email me at josh@ecommbreakthrough.com and in your subject line say “strategy audit” for the chance to win a $10,000 comprehensive business strategy audit at no cost!

Transcript Area:

Andri Sadlak 00:00:00  The shift that we’re talking about is search engines are becoming answer engines plus action engines. So it’s already answer engines. We’re already deep in it. Most of the people at least. And action engines is what a lot of people predict is going to happen next because we’re going to save our time. We’re going to go for convenience and trust AI. We already do, right? The good news you’re watching this. So now you know that you need to pay attention. And I’m going to share with you exactly what I need to do.

MC 00:00:29  Welcome to the Econ Breakthrough Podcast. Are you ready to unlock the full potential and growth in your business? You’ve already crossed seven figures in sales, but the challenge is knowing how to take your business to the next level.

Josh Hadley 00:00:43  The biggest opportunity in ecommerce right now isn’t a new platform. It’s actually a brand new buyer. AI agents that will choose products on behalf of customers, and the brands that figure this out early are going to have a massive advantage. Welcome to the Econ Breakthrough podcast.

Josh Hadley 00:00:58  I’m your host, Josh Hadley. I scaled my own brand from 0 to 8 figures in sales, and now my mission is to take it to over nine figures on my journey to nine figures, I bring you unfiltered conversations with the smartest minds in the e-commerce space. Past guests include Ezra Firestone, Kevin King, and Michael Gerber, author of The Myth. Today, I am going to be sharing with you a fascinating conversation with Andri Sadlak. He is somebody that I met again last week at the prosper show, and he had a fantastic presentation that was all about agent commerce, how to get optimized for Rufus, and how to really get your brand and your products visible for this next wave of agent commerce. Andri Sadlak is a serial entrepreneur and founding head of product and strategy at ASM for building AI first e-commerce software for the age of agent shopping. He’s been an Amazon seller since 2017. He exited his own brand and co-founded product pinyon. Today, he’s part of the team behind Amazon growth and AI visibility strategies for eight and nine figure brands like Mars, Oriel and HP.

Josh Hadley 00:02:02  He’s one of the sharpest voices in agent commerce, taking the world stage to break down how brands need to show up in the age of AI. Shopping assistants like Amazon, Rufus, Walmart, Sparky and Lem search. With that introduction, welcome to the show, Andri.

Andri Sadlak 00:02:16  Thank you so much, Josh. Real pleasure being here.

Josh Hadley 00:02:18  I’m super stoked to have you on the show. And I know our listeners are going to be very excited. What you shared was and I rarely do this. Rarely do I see, like a presentation at a conference that I’m like, that was so good. I want you to come back and like, literally regurgitate the same information back to the entire podcast audience. But I genuinely feel like everything that you touched on was, to me, something that I took religious notes on, that I shared with my own team that I felt like, look, if I’m doing that, this needs to go out to everybody listening to this podcast. So, Andri, you did a great job on that presentation, and I’m excited to have you share that with the rest of the world.

Josh Hadley 00:02:54  So without further ado, let’s dive in to your presentation.

Andri Sadlak 00:02:58  Yeah, that’s a very high price praise coming from you, Josh. And before I dive into the presentation itself, quick disclaimer everything useful that you’ll see there comes from my clients, from my team, and from the sellers I had the pleasure of working alongside since 2017 and all the bad ideas, they’re probably mine. Okay. And if you take anything from what I learned so far and what I’m going to share, I wanted to take two things. One, I want you to use it for good. Don’t weaponize it against your competitors. Don’t deceive your shoppers. Okay. And number two, pay it forward. Pass it on to another ethical brand builder. So none of the knowledge dies in. We want this to be used and want our industry to a level up and again made plenty of mistakes. You even have to sell my house at some point to save my business. So if you do see something wrong, speak up or comment wherever you see this and we’ll all just leave smarter.

Andri Sadlak 00:03:46  And that’s the main goal.

Josh Hadley 00:03:47  Fantastic. Well, as you pull up the slides here, Andri. Yes, I think what he touched on was super valuable. I think like one of the best ways you could pay it forward to other people is genuinely sharing the show. So if you get value from today’s episode, make sure you go drop it in a slack community, a Facebook group that you’re a part of, whatever masterminds like. If you got value from this, share it there. If you were just listening to the audio version of this, make sure you come check out the YouTube version of the podcast so that you can see the slides. What I love about Andri is like, this isn’t just theoretical stuff that he’s going to be walking you through. It’s very tactile to say, go navigate over here to Amazon. This is what it looks like. Go take this. This is how you go spin this up. So Andri you got your slides showing up here. Let’s dive in.

Andri Sadlak 00:04:31  Sweet.

Andri Sadlak 00:04:32  So the top is in the e-commerce three levers to boost your brand’s AI visibility. And here’s a big question. What if your next customer never search for your product. Never read your listing and still above from you. Sounds cool right? My name is Andri Sadlak. Monticello since 2017. Sold my brand in 2020. Co-founded Product Pinion scale seven and eight figure brands as a consultant, as well as an operator in an aggregator, and now in building a Azoma, which is an agentic commerce optimization software. Azoma is basically the right hand to some of the biggest brands you may even have at home, in optimizing their visibility in the new world of AI to work with Mars, P&G, Unilever, Beiersdorf, L’Oreal, Colgate, HP, and so on and so forth. My promise to you is by the end of this presentation, you’ll have some actionable items you can execute on or pass on to your team. That’s always my goal. I hate presentations that just talk about fun stuff, and it all sounds amazing, but there’s nothing you can really do today.

Andri Sadlak 00:05:25  And my goal is to change that. And let’s dive into it. I’m going to break it down into three parts. First the shift. Why now owns the shelf. So a little bit of explanation. The background of what’s happening then three Amazon levels, the science and your actual action plan and if time permits will open up some Q&A. So I’ll discuss with Josh and obviously feel free to hit me up afterwards as well. So let’s dive into it. The shift why AI now owns the shelf. Sounds futuristic right? Now quick question. Do you guys use ChatGPT in the audience? Pretty much everybody raise their hand. The second question is, do you use GPT? Decide what products to buy. Do you judge by any chance? Do you research products?

Josh Hadley 00:06:01  So actually, it’s funny that you ask this because literally last weekend our dishwasher went out and like the first place I went was ChatGPT and I was like, what is the best dishwasher? And I was like, hey, imagine that price is not like not even in the equation.

Josh Hadley 00:06:16  Like genuinely like, give me the best of the best. And then obviously I was able to like hone it in from there. I was like, yeah, I’m not buying a $20,000 dishwasher, but like help narrow it in. So I this is definitely where the world is moving. Totally agree.

Andri Sadlak 00:06:28  Yes. And this is a classic example. I’d say pretty much everyone I know uses ChatGPT for product research at this point. Another question do you shop on Amazon? I would imagine pretty much all of the listeners would say yes. And have you actually tried, Rufus, to decide what products to buy? Have you Josh?

Josh Hadley 00:06:44  Yeah. So I actually personally have not I think I typed maybe one question into Rufus because I was like, how good is this thing? But then after again, your presentation, I actually spoke with my parents and I was like, have you guys used this feature? And they both were like, oh yeah, we’ve we’ve dabbled in this. And so like it’s like 5050 from like again, my parents, my current usage.

Josh Hadley 00:07:05  But I know the data that you have. I was actually blown away like the number of people using Rufus. So it the wave is coming is just a matter of time until it’s like fully adopted. But I think long gone are the days of people just typing in like keywords into the search feature of Amazon.

Andri Sadlak 00:07:21  Yeah. You know what’s funny? They built it so well into Amazon. And sometimes you don’t even realize that you use Rufus. For example, Rufus works with text. You can just type obviously questions voice. A lot of people just like voice messages. We all have that friend who sends us that long voice message. And Rufus works the same way for us. So you can just take a picture of the product. And the cool part is also has price history, so you can even task. Rufus, let me know when the price drops, right? My favorite is definitely taking a photo. I just took a picture of my backpack because now I need to buy a new one. I wanted to find the exact same.

Andri Sadlak 00:07:48  I’m very boring and easily identify what exact product that is, so I don’t even have to look for the name what the exact model is right now. The cool part is it has your purchase history so it knows you really well. As a shopper may not know you as a person as well as ChatGPT would if you talk about your life issues with ChatGPT, but your shopping history that’s Amazon Amazon’s advantage, right? And Andy Jassy, the CEO of Amazon, is publicly very loud and proud that he is bullish on a giant e-commerce. This is a video from Davos. If you look it up you’ll see what he’s saying there. But basically he’s saying not nobody else has the buying history. And that’s our advantage. And rightfully so. He’s bullish because Black Friday last year just a few months ago AI tools influence 14.2 billion in global sales billion with AB3 billion only in the US, which is 805% year over year growth. 38 to 40% of Amazon sessions used Rufus during Black Friday. Pretty insane, if you ask me.

Andri Sadlak 00:08:42  So Rufus is basically Amazon’s gateway to a genetic commerce and what exactly genetic commerce is. That’s when you don’t search what AI searches, compares, filters and buys for you. Sounds pretty futuristic, right? And if we break it down into the exact steps, first we start with the intent, right? We have an issue to solve. Then we get recommendations. And then we have an option to transact. And in the old old world, the world all of the sellers are familiar with. We used to break down that customer journey into search. Fine. By right. So the idea of genetic commerce is all of this will be done by the agent. The big question is are we there yet? And realistically completing the purchase in that answer engine. So any of the agents you’re using like ChatGPT, perplexity, Gemini, whatever you like, including Rufus is currently the least adopted use case. But it’s coming, right? It’s in development. So if we look at the biggest players, OpenAI develop this big new protocol.

Andri Sadlak 00:09:33  It’s called a genetic commerce protocol ACP with intact checkout. Right. They were the first two to announce that they’re working on it, but they pulled back six months later. Now they’re restructuring. They changed some big things. Everything’s kind of confidential, but they’re still working on it. But it’s not going as smooth. Google developed a similar thing called UCP Universal Commerce Protocol with over 20 partners that basically provides live checkout as well inside that AI mode and Gemini now as well. Alibaba in China is actually the only fully closed loop because they own the AI, they own the marketplace, the payments and the logistics. I think the reason the Western systems are a little slower is they want to partner with existing players. In Alibaba’s case, they just own the whole infrastructure, right. So the most generic commerce today is still mostly discovery and assisted checkout. So human is still in the loop. It confirms. But it’s for now. And if anyone’s selling on Shopify they probably go. This email I got it I think March 10th or 11th just before prosper, they basically says, hey, Shopify is here.

Andri Sadlak 00:10:30  Just so you know, moving forward, your products will be searchable on ChatGPT. You will be able to find them and buy them. So there’s definitely progress. It’s happening behind the scenes. If we’re deep into the grind of Amazon, we may not notice how fast it’s developing, but I’ll show you some data so you understand the importance of paying attention. We’re honestly at a similar inflection point to the early 2000. So if you look at this graph, if you’re watching it, you’ll see that ChatGPT adoption skyrocketed recently. So if you compare it to X, Instagram, Facebook or TikTok, you’ll see that it took them way longer to get to a billion users. We’re almost at a billion users with ChatGPT in about three years, so quite insane. This is really the fastest growing product we’ve ever seen in the history of tech. Now, if you compare ChatGPT to mobile phones actually to yeah, to mobile phones, it took 20 years for mobile phones to get to 1 billion users ChatGPT only three years or so, but llms as a whole.

Andri Sadlak 00:11:18  So all of the agents that that we use. So LLM stands for large language models. 20 years to 1 billion with mobile phones but ChatGPT Gemini, meta AI, deep sea perplexity, grok copilot cloud hugging, face manners, all of that together 2.3 billion users. That’s one third of the world’s population. Pretty impressive right?

Josh Hadley 00:11:38  Yeah. So that’s that’s what’s that’s what’s mind blowing about this is how quick the change is coming about. And I think it’s not even this like, oh well, I’m gonna wait for a few years and take action on this. This is like you got to be changing and adopting this in real time, where we I think a lot of us grew up in the space of, like the internet coming of age and social platforms, coming of age. And it was like, hey, have you adopted Facebook five years into, you know, it’s launch? Like you were still kind of early on. Right. But I think that’s what’s changing is like the speed at which things are transitioning is super fast, which is why I think, like what you’re sharing is, like so poignant today, because this is not something you can sleep on.

Josh Hadley 00:12:18  If you do sleep on it for a period of a couple of years, by the time you’re able to adopt it and change and adopt any of these strategies, you’ve already lost so much ground in market share that it may be extremely challenging to recover from 100%.

Andri Sadlak 00:12:33  There’s definitely advantage of being an early adopter, and that’s basically who we’re talking. So back to some numbers. 3 million daily average visits. So not users yet across all of these lamps right. Impressive. ChatGPT 75% of that. Still Gemini is on the rise. So if we compare the traffic ChatGPT is now the number two for discovery platform right after Amazon. So basically half of it. But think about adding to it. Gemini Cloud Copilot perplexity. Anybody else is going to come into the game. It’s probably going to be very similar to the traffic on Amazon, so cannot be ignored. Another study from Accenture. They basically ask people what is your preferred source for purchase recommendations? And consumers basically trust the AI much more than friends and family, social media and marketplaces.

Andri Sadlak 00:13:13  So it’s already the second highest trusted source after physical store related to physical humans helping them make decisions. Now, Amazon and OpenAI, I don’t know if anyone’s paid attention, but I definitely do. Amazon invested 50 billion into OpenAI. That’s that’s a huge number. It’s a big deal and it may change a lot. And Amazon is actually already using their technology. So OpenAI models are used across Amazon’s AI products and agents that serve customers directly. So that AI.

Josh Hadley 00:13:40  Real quick on that note, but it was interesting that Amazon invested so much into OpenAI after basically telling OpenAI, like, by the way, do not crawl our Amazon listings and do not offer up our Amazon listings in your LLM. So like what? Why like why the change of heart? Like what do you think? I know it’s anybody’s guess. But like what do you think is going on here. Because like I can’t imagine Amazon like losing a bunch of market share to, you know, these llms that only recommend Shopify products or something like that.

Andri Sadlak 00:14:09  Yeah, exactly. That’s exactly what’s happening. Don’t want to lose the control. That’s why they invested while still protecting their data. So they’re not going to share their shopper history. Whatever purchases have been made with OpenAI, I don’t think at least. But they will use AI, OpenAI technology to make Amazon’s experience even better. Right. So what I think what they’re trying to do here, that’s again, speculation is use the existing technology. People clearly allow fastest product in the world, right? And enhance that with with that product and hence Amazon right. So Amazon will still own the customer data. But the way it’s it’s working will probably be more like OpenAI. It already is very similar. But I think they’re going to adopt it even more so that AI is actually reading your listings on Amazon today. Amazon’s controlled OpenAI part of the deal, right? So we can’t just ignore it. Rufous conversion if you compare it to non rufous conversion is already incredible. So this data comes from Sensor Tower. They just show the graphs.

Andri Sadlak 00:15:04  You don’t see the air. And basically no rufous conversion 11% rufous conversion roughly 36% during Black Friday last year. So 3.5 x improvement when you can actually ask questions and get the answers right. And in 2025 over 300 million customers use. This comes from Amazon’s Q4 report, and the JSE was very proud to announce that monthly account users went up by 149% year over year, and interactions went up by 210%, so more than they expected. Another big part he keeps saying to all the media customers are 60% more likely to complete a purchase. So if you talk to roofers, more likely they’re going to buy. And number three, Rufus generated 12 billion in incremental annual sales. So they projected to reach 10,000,000,000 in 2025. They crossed that. So definitely exceeding anybody’s expectations. And again not everybody understands they use Rufus when they just especially on the mobile phone I think it’s integrated so well. Now if you’re selling in pet care or travel or entertainment, pet care is the best. But Traveling Salesman follows quite closely.

Andri Sadlak 00:16:05  This is these are the spaces where shoppers tend to use lambs the most. So you said Joshua, resource your dishwasher or whatever other product you were looking for. A lot of people research backyard or travel and entertainment. Obviously all of these spaces get some traffic, but these are the biggest. And that’s based on Bain and Company’s research of how frequently gen and AI is being used by shoppers in the US. Another interesting point here is US consumers said they would trust to automate purchases with AI, their familiar marketplaces. So not necessarily tech companies or AI focused companies, but the marketplaces, which again, emphasizes the point that optimize on Amazon for genetic commerce makes a lot of sense because people tend to be more likely to agree, hey, buy whatever I bought last week or find that exact same mouse and order it again, you know?

Josh Hadley 00:16:48  So Andri, I just had like kind of like an unlocked mental unlock here, as you mentioned, that I think that what Amazon is known for is being one of the most customer centric, like companies out there.

Josh Hadley 00:16:58  They pride themselves on that. Right? With the adoption of AI. I truly do believe that, like if somebody is going if a customer, even myself, like I haven’t introduced a genetic shopping into my life yet, I’m sure it will happen. But what better platform than Amazon? Honestly, because it knows all of my purchase history. And guess what? If something shows up, I’m like, what the heck is this? Oh, it’s so easy to return. And I think Amazon’s going to make that so seamless to where it’s like, I am going to go reorder products for you on autopilot, and I am going to you tell me a kid’s birthday’s coming up, I’m going to go order that product for you, and it’s going to show up, and you’re either going to love it, and if you don’t like it, then just return it. It’s super simple. That I think, is the unlock with Amazon. And I think that Amazon gets the AI agent ecommerce shopping done fully, autonomously, faster than even open AI, I would argue because like, who wants to go through like the return experience of like 500 different Shopify brands, right? Like, that’s not fun.

Andri Sadlak 00:17:58  Exactly. Yeah. Well, the promise that they made with OpenAI, to be fair, is you’ll just deal with us directly and we hold the whole journey. But what’s cool about Rufus, as you said, it’s personalized. So even the search results or I should say results when I’m asking was the best, I don’t know, key helmet. And when you ask the same question, we might get different results based on our purchase history. That’s how good it already is. And it used to be super fun in the beginning. Now it’s getting really smart. Okay, now let me show you some of the fun stuff in terms of businesses. This is a study from NGP digital. This is class lots of businesses. How well they use generative AI optimization. Basically small businesses are the least adopting category of AI optimization. Medium businesses are the best so far. 21% of them actually adopt GE optimization, compared to 8% with small businesses. And enterprises are kind of in the middle 16% of the enterprise level businesses already optimized for Geo.

Andri Sadlak 00:18:52  And that’s the target audience Asamoah is working with. And it’s kind of fun to work with them because the reason they are number one still, after being so like being the best in the game for so many years and selling the most is they actually have budgets for whatever new comes in and they dive into it right away as soon as they can. Now the shift that we’re talking about is search engines are becoming answer engines plus action engines. So it’s already answer engines. We’re already deep in it. Most of the people at least. And action engines is what a lot of people predict is going to happen next. Because we’re going to save our time. We’re going to go for convenience and trust AI. We already do, right? The good news. You’re watching this. So now you know that you need to pay attention. And I’m going to share with you exactly what I need to do. So as an Amazon seller, the big question probably have in your mind is like, so what? What do I do? Right.

Andri Sadlak 00:19:37  And the next chapter of what I have to share is the three Amazon levels, or the science behind Amazon’s AI and your actual action plan. What exactly do you need to execute on? Ready to dive in?

Josh Hadley 00:19:49  Let’s do.

Andri Sadlak 00:19:50  It. Let’s do it. So quick story. I cook once a year. My wife’s probably happy that I don’t cook much more often because I’m not the best cook. Self-awareness. And when I asked what is the best pasta for romantic homemade dinner on ruthless. Last time I saw these results. So if you see the left side of the presentation, you’ll see the ruthless results. These are the products recommended. And then I also searched for dry pasta. So on the right side you’ll see the traditional search results where all used to. And I was quite surprised that I was scrolling through it and none of the results matched up. Left and right. Let’s remain different. Even the top rated. The best sellers. All of the main, main products with the biggest numbers of reviews were not recommended by Rufus.

Andri Sadlak 00:20:31  I kept scrolling, left and right side were very different. So the big question that I have to you listening is which part of the screen you think will see more development? The left side Rufus or the right side the traditional search. And my personal bad is definitely the left side. And if you’re with me, the big question we all have is how do we show up in Rufus recommendations much more often? How do you optimize for it? Right. And I have three levers to cover with you. Lever number one is Cosmo the knowledge layer. Lever number two is Rufus. The conversation layer. And lever number three is citations the trust layer. So let’s.

Josh Hadley 00:21:04  Dive. I think that like the the scary part that you just shared is that not even the best sellers are being recommended in Rufus, which means like if you’re sitting there as like a category, like you’re just dominating the category and you have been for the longest time, I think, like we’re about to welcome in an age over the next couple of years where there can be massive disruption like we may be seeing like a period of time over the next, like I would say, probably like 18 to 24 months on Amazon as like Rufus figures out what it wants to serve to people that like, it’s kind of like launching a new product in 2015 on Amazon.

Josh Hadley 00:21:39  We all wish we could have those days again, but like you now have that opportunity to, like, upset a best seller if you execute this properly. And if you’re a best seller to maintain your best seller status, these are the actions that you need to be taking. Otherwise, a new entrant or an existing competitor can go implement these things and start gaining market share on you. And you’re like, you have no idea where it’s coming from, but it’s because of the way agent e-commerce is shifting on Amazon.

Andri Sadlak 00:22:06  So well said, Josh, I appreciate it underscoring that, because what we’ve seen in the comparison here, Rufus, versus traditional search, clearly visualizes that it’s two different algorithms that Amazon is using. And obviously we don’t depend on one fully at this point. And the only question is which one do we think is going to get developed more and where the pack is moving right. So people how will people shop moving forward? And this part of my presentation is basically focused on the future, on future proofing your business.

Andri Sadlak 00:22:34  And if we look back at the history so something referred to like 2010 to 2016 back then that that era that I call listings 1.0 was basically keyword games, right? Keyword stuffing review and rank manipulations. And if you’ve been selling that time you’re listening to this, probably you can probably relate. I started researching Amazon in 2016, so kind of late to the game, but launched in 2017, 2017 to 2024. I call this year listings 2.0 content. The brand is like how you win better content. Better reviews lead to higher click through rate, higher conversion. That’s how you went. Right now, 2025 onwards is the era I call listings 3.0 AI visibility time. So that’s causal understanding rufous answering questions which leads to better conversion to and external citations validating the responses. That’s the era we’re in right now, in my opinion, and that’s what we will focus on. But I also want to emphasize listening to people still exists. People still search traditionally, so don’t ignore that at least yet. These two approaches coexist right now and have to be taken into account when optimizing, moving forward.

Andri Sadlak 00:23:36  And now the exciting thing that was just announced, like March 10th, right before prosper, I got a message from the Amazon employee who I met at one of the events, and she basically said, the email is going to come out soon, but I know we post about this stuff. Here you are sponsored prompt prompts and sponsored Brands prompts are going live on March 25th. So if you’re listening to this, after that, you’ll know this was all in the works from like fall 2025. Amazon invested a lot of resources into developing this. And this basically what it looks like. You’re going to scroll through the listing. You’ll see a field that looks like suggested Rufus. Questions. Hey, this is what you should ask. One of them will just say sponsored. I already see this. Some of the listeners may already know what I’m talking about. If you click on that sponsored one. It’s going to show up with an answer, right? And the way the Amazon staff explained that to us. We met in Milan, actually in November last year for Amazon Ads Creators Retreat, and they kind of used us as a as a playground and as a brainstorming session and a focus group on how to develop this further.

Andri Sadlak 00:24:28  What they explain to us is we want this to become like a paid employee in a department store by the brand. So, you know, if you walk into the department store, there might be someone at the entrance kind of directing you where to go and making sure you’re happy, smiling and say nice things so you spend as much as you possibly can. But then there’s another employee. So the first example is Rufus. Right? Then then there’s another employee. If you walk into a store and you ask questions about specific products, that will be apparently by that brand selling those products so that sponsor prompts in their worldview. So this is an example of a sponsor prompts report for anyone listening. You don’t see this. It looks very similar to any traditional ads report. And it’s still in beta for now. It probably won’t be when you listen to this, but you’ll see like click through rate CPC, spend your actual sales. Write for now. Show zeros in the reports I work with because it’s beta, but it will become paid from the 25th of March 2026.

Andri Sadlak 00:25:20  Ads console campaign ad group ads and prompts tab. That’s where you find it. Definitely look into it. You might even already have this active without knowing, right? Because it’s free and they automatically enrolled on most of the brands. So what we do know it’s coming live 2025. Sorry, March 25th, 2026. So you’re going to be paying for the clicks. Amazon writes a prompt and the responses you write nothing, right? You have no input in terms of how to respond to the questions. So your listing actually becomes the ad creative. Amazon will write the prompt and the response so you don’t have any say there at all. Your listing will basically become the content that is used in the ad creative. So your weak PDBs will become weak prompts and you will pay cost per click for that. Answers right. So number one thing you need to do is find the report audit it and pause. Pause any individual prompts in Ads Console. So whenever you start paying, and you probably are by now, you don’t pay for something that does not get any clicks.

Andri Sadlak 00:26:13  Okay. What are your thoughts on this problem so far? Josh?

Josh Hadley 00:26:16  Yeah. I love the sponsored prompts. I actually messaged my team after the presentation. I was like, hey, are we getting a download? Like, what are people asking about on this? So now that we know we’re going to have to be paying for these placements, it’s going to be even more important to be optimizing these and making sure we’re on the lookout for.

Andri Sadlak 00:26:32  It 100%, 100%. Very important now that Amazon is throwing money into this. And that’s how they make money, right? Ads. We know that they’re serious about this shift. So sponsor prompts are basically made of PDP brand store and keywords from existing sponsored products and sponsored brand campaigns. So if you look at the levers that we’re impacting, Cosmo and Rufus, the knowledge layer, and then Rufus the conversation layer, it’s basically what Rufus knows and what Rufus says. Sponsor prompts will just be a paid amplifier, right? So we pay Amazon to say everything louder.

Andri Sadlak 00:27:00  Okay. Now, number one, let’s focus on the actual actions we can do to optimize. Lever number one is Cosmo the knowledge layer. right? So you probably know something about cosmos by now. Cosmos is one of the Amazon’s algorithms, and the way we optimize for it is some basics and some advanced stuff. Number one, browse node and item type. Integrity has to be right. Number two backend taxonomy has to be optimized. And number three there are 15 relation types that may apply to your product. And we also need to write everything in noun phrase way. So noun phrase optimization is what NPO stands for. Now number one let’s focus on how do we do the basics right. So browse node is actually important. A lot of the sellers have been playing with this, moving their products around to increase their chances for getting number on best seller. You have to be very careful now, because if you are in an incorrect browse node, you’re just not going to be understood fully by cosmos and recommended by Rufus as a result, right? So make sure your browser node is correct.

Andri Sadlak 00:27:54  And in the back end at the very top you’ll see one of the backend attributes. Same item type keyword. That’s basically explaining what the product is. Make sure this is done correctly as well. Okay. Number three.

Josh Hadley 00:28:04  How do I make sure I’m in the right product.

Andri Sadlak 00:28:05  Type? You’ll have to do a little bit of digging in my opinion, using an LLM, ChatGPT or Gemini or a clod. It’s probably my favorite. And throw some competitors into the into the chat and ask what’s the right item type keyword for my product? I’m selling this in that that’s what I would do because I don’t know the exact product of selling, but sometimes people just intentionally skew it just to get into a different competitive landscape. In the Amazon algorithm, this is not going to help you anymore. Okay. So number two, back end taxonomy. Basically the back end attributes you need to optimize it. And I created this a little skill. So think of the skill for Claude as a prompt you can upload right I’ll share all of these files at the end through a QR code.

Andri Sadlak 00:28:41  So don’t worry about this. But all you have to do is basically say to Claude, use this skill, name the skill to optimize the backend attributes for cosmos and the way I do it. Because I’m pretty basic, I’m not that tech savvy. I just upload a printout of my backend, specifically product details that has all of the important backend attributes into that chat. But some people also who are more advanced, they understand flat files better. They can upload flat file and get similar results. So all we have to do is upload the existing situation, use that skill that will analyze the existing backend attributes and give you an action plan like this. It’s going to tell you here’s the attribute name. This is the status. It’s weak or something’s missing. And then this is the current value. This is why it matters. This is why we recommend optimization. And here’s an example of what can be like right. Examples are quite often using this skill what you can use. But you still need to check to make sure AI is not hallucinating, right? Yeah.

Andri Sadlak 00:29:34  So this is kind of my process so far outside of some of course ASM has this built in into the software tools, but I’ll share my skill with you so you can use it in the future. And obviously you can edit the skill if you find a better way of doing this. Okay, now the third part, and this is the fundamental part of cosmo optimization. There are 15 relation types. And if you scan the QR code if you’re watching this. So if you just Google Cosmo paper, you’ll probably find the link to your PDF that Amazon kindly enough share with everybody publicly. And you can read that. And I encourage you to do so to understand how the whole causal algorithm is built. Right. And based on that, we can also optimize your whole PDP. So as an example of what they have in the whitepaper, they use this pregnant woman shopping for shoes as an explanation of how the algorithm works. Essentially, the pregnant woman looking for shoes ends up buying slip resistant shoes. Cosmos stands this user behavior and claims that a common SaaS relation that pregnant women require slip resistant shoes, right? So we kind of want to help Cosmo understand these things by providing all the necessary information about our products.

Andri Sadlak 00:30:32  For example, if my shoes wouldn’t be specified as slip resistant, I probably wouldn’t show up for this pregnant woman talking to an agent. Right. And in short, there are 15 relation types that may apply to your product. Not all of them do usually, but these are all of the ways you can describe the product, right? It’s basically the function, the event, the audience, the what? What is the product type? What is the body part? You apply this to anything that it works with as a complementary product and so on and so forth. Right. There are 15 of them. They also gave you examples in the white paper they can go through. But what we have to do is basically look at our products, see what applies. So, so far from what you’ve seen. Josh, I know you haven’t been a big fan of rufous yet. Do you understand the difference between Cosmo and Rufous?

Josh Hadley 00:31:12  Yeah, 100%.

Andri Sadlak 00:31:14  Okay.

Josh Hadley 00:31:14  And that’s that’s the important thing is like us or those listeners, like, there are two different, like, algorithms that you’re kind of like optimizing towards or knowledge bases.

Josh Hadley 00:31:23  And that’s the fundamental thing you have to learn.

Andri Sadlak 00:31:25  Exactly. Yeah. So to summarize for everybody else, Cosmo Rufous Cosmos, the common sense algorithm basically answers the question, what is this product, who it’s for, why people buy it. Right. And Rufus is a shop assistant. So which product answers your question? Your as a shopper, right. And in more technical way, Cosmo is a product meaning text attribute extraction, use case mapping, common sense tagging and Rufus is about intent matching synthesizing all the reviews and QAS visual label tagging. So Rufus is the one looking at images and then answering and ranking the products for you, and together they work to align the product and the intent. Okay, so lever number two is Rufus the conversation layer. So let’s see how we can optimize for Rufus. A lot of people by now have tried it. Some love it, some hate it. But Amazon’s Rufus is the way Amazon is entering the e-commerce game, and it’s the shopping assistant that’s going to dominate the new shopping behavior on the platform.

Andri Sadlak 00:32:21  The number one thing we can optimize is, well, the step we need to take is the audit. So we need to collect niche questions and understand how we actually answer that. When people go to our listing and ask Rufus. Right. So we test those answers and our listings. Number two, we’re going to enhance the copy the images and the A+ to answer those top most important questions people tend to ask in the category. And number three, we need to seed Q&A to enhance the answers even more. Okay, I’ll dive into the details. Number one, the way we audit simply go to listings. You see these suggestions right below the image and right about their views. And you can see what people tend to ask is Rufus shows what shows up most often. Okay. If you look at the slides I’m sharing, you’ll see on the left a screenshot of a report. So this report comes from the report comes from Azoma where we basically show how many appearances every question got in the whole niche.

Andri Sadlak 00:33:10  Right. So if you if you can see the screen, you’ll see can it be used daily? It appeared 447 times I believe that’s in the last month from the number of times we collected. So 81% is the average score of the answers, meaning how well the answers come in in those listings. And this way we can find opportunities to optimize for. Right. And not all the questions you’ll see on your listing will be the exact same on your competitor listings. We want the data on the whole niche okay so far. Makes sense.

Josh Hadley 00:33:34  O makes a lot of sense. This is not just your listing. Make sure you’re doing this across your competitors because they’re going to be asking different questions. So love it. Big takeaway.

Andri Sadlak 00:33:43  Amazing. Yeah let’s move on. Quick tip if you are testing these answers in your listing and you see that the answer from Rufus comes with something like most customers report seeing results, whatever most customers are referring to the customers. That means that Rufus is looking into the reviews and the Q&A section to answer this question, which also tells you, hey, I can optimize this.

Andri Sadlak 00:34:02  It’s clearly not using my inputs on the listing to answer the question, right? Some of the questions may not be something you completely can answer, and that’s what we’ll cover in the step number three of the Rufus Lever optimization. Okay. Now let’s go into the number two, the enhanced part. So the way we enhance copy images a plus is is very straightforward. But I also build a skill file basically a prompt you can use in cloud by just uploading my file. That helps you answer the key Rufus questions in the copy. So let’s start with the copy. So you can use this prompt that I shared with you in the slides and upload that skill to basically create a new, better title bullet description. Okay. The way it’s doing it, it’s using Cosmo and Rufus data that you upload. So you need to upload the Rufus questions you collected, and it’s rewriting it slightly by keeping as much as the original as possible not to mess up your SEO. As an example. The output will be a table with a current version, with the suggested version and the science explanation on the right in the column.

Andri Sadlak 00:34:58  Okay, now the second part images. Images are critical for reviews. Okay. It’s not something we can ignore. Has never been the case really on Amazon. Now it’s not just for humans but also for the AI. So we want to front load key information by reordering the images. That’s the first step even for humans. Normally people don’t go through all of the images right away. They look at the top one, two, three and they make the decision, do I need to start this further? Do I skip and look at the competitors or do I buy already? Right. Same for AI. You need to put what’s the most important at the very top right. So there’s a quick skill. Another file you can upload to cloud that will reorganize based on the reviews questions you upload it to maximize the chances of answering the questions with images. And the output will be something like this. Current order. Suggested order. So new order of the images name by the type of the image and in the bracket will see the previous position.

Andri Sadlak 00:35:44  Okay, obviously review this manually as well before you make any big changes, especially if it’s one of the best sellers to make sure it all makes make sense now.

Josh Hadley 00:35:51  Andri. Real quick with the images. And I know you’re going to share an example here, but I’m almost wondering, like, does it make sense to like, take a copy of, like, the most frequently asked questions across the niche and like, almost like add an image that is just like FAQ based only. Do you think that Amazon’s reading that image and extracting all of that? Or is it like two different things like cosmos not pulling it so Rufus isn’t really getting it?

Andri Sadlak 00:36:16  I love.

Josh Hadley 00:36:16  The question. Yeah. What do you see on that side?

Andri Sadlak 00:36:19  Yeah. And the asset comes in right after this, so I’ll dive into it. An image really is worth a thousand words, especially if it’s an image like this. Right. So here’s an image that was shown by Rufus to me a lot when I was looking for a clicker.

Andri Sadlak 00:36:30  When I was asking something about the connectivity, the charging. This image would be shown by Rufus as an answer with some text, but you know when image is shown, the conversion goes out by a lot. And the way Rufus is working, it’s using multi-modal understanding, right? So all these images are being scanned for text objects. The people in the image use cases. The actual context of the use cases right to match shopper intent and generate accurate answers. So images are scanned. And there’s another technology by Amazon called Amazon Rekognition that sellers can also access the register in AWS and look through it and find it. And you can upload images that way and see what Amazon’s AI recognizes, right? And as an example, here’s a screenshot as well. The question I ask from Rufus is is this shampoo good for curly hair? The answer came in properly. There’s plenty of information. Thanks God. This listing is quite well optimized for this question, but what’s interesting? The image also was shown by Rufus.

Andri Sadlak 00:37:22  In the image, they’re showing a person with curly hair and they’re looking at plenty of images. This is just the correct match for the question I asked. Right. So we know Rufus is actually understanding what’s in the images.

Josh Hadley 00:37:32  Yeah I think that’s a mind like that’s a mind blowing change where like Amazon images have never been part of like the SEO game and like the traditional Amazon aspect, there’s things that you could do with A+ content and alt text and things like that. But now like it’s actually reading images. So it makes, yes, like your images ten times more valuable and important, which again, I know there’s a lot of people just using AI to spin up images. I would just be very careful, like you could do that, but make sure it has context of like what the top questions are so that you actually get served up here in these questions on Rufus 100%.

Andri Sadlak 00:38:08  Yeah. And the absolute lowest hanging fruit, if you want to optimize optimized images, is adding the right text callouts on the images.

Andri Sadlak 00:38:14  They actually answer the Rufus questions. Yeah. Simple way to do this. If you don’t want to create that AWS account and upload images, when you add a product on your cell central, you can upload the image and you can click this button Generate content. Okay. When you do that, Amazon will create a title based on what was recognized and understood from that image. It’s just one last check you can do to make sure there’s nothing that comes with a red flag that can have an issue with compliance, but also gives you an idea of what is really understood. So in this case, it was a clicker. Amazon thought it was a portable power bank. So what I would probably do if I was selling this, I would probably add a packaging somewhere on the image that has the product name. So at least Amazon can recognize the text and see what product that is. So quick. Quick check like that will help. Now speaking of images again, we want to maximize what Amazon is calling OCR optical character recognition.

Andri Sadlak 00:39:04  So answering questions in the image, even if you don’t want to change the images themselves, add some clouds. Right. Just like I said. And you can use the same chat you started with cloud already. Add this new prompt that I’m showing on the screen. Share, and that’ll probably give you some good ideas of what other questions you can answer in your images, and which exact images to use for that. So as an example of that workflow, you’ll get the images on the left, the additions of the text on the right, and then further right in the column. Exact questions we’re answering from the questions you uploaded, the questions that we have in the workflow. And my favorite would be the last example here is basically white canvas just showing some kind of shampoo parts. And Claude came up with so many ways we can answer questions that Rufus has in that image just by adding some text, and these are easily low hanging fruit for image optimizations. All right. Now alt text. So unfortunately alt text only exists for A+.

Andri Sadlak 00:39:55  And we’re going to optimize a plus alt text for images. So every image on a plus will have an edit button that you can click and add a honey character long explanation of what is on the image. So another good way to optimize it. Definitely do that. I normally just download it, throw it to Claude and ask for some help. How do we answer these questions best? Because it already has the Rufus questions in the workflow, and it gives me some ideas for how to write out the alt text. And now the thing that I recommend everybody do in a plus, there’s a section you can add. Some of you may have it which is Q&A right. So very straightforward and big picture LMS. These large language models, they’re all about questions and answers right. So add that section if you haven’t already and find a way to create the questions and answers based on the Rufus data. What people tend to ask, you just have to phrase it in a brand coherent way. Provide those answers in there.

Andri Sadlak 00:40:44  So there’s another quick prompt you can use to write it out properly. And I would again manually review it just in case to make sure it fits your brand guidelines. This is what it looks like. So very simple. You open up the answers and there’s no reason to believe Amazon didn’t create this for a reason. And we got to use what Amazon gives us. Okay. Third one. This is a very funny part that like our enterprise level clients are very careful about. Most of the sellers know this. So I’m not going to repeat this too many times. But there is a section right about the reviews on the listing where if you click on one of the questions suggested, you will see there’s a line here. Show related customer reviews in Q&A. If you click on it, it’ll open up the actual reviews and Q&A is used by Rufus to generate the above answer to the question you just asked. Right. So the way we can see this is if we click on click on the answers, it’ll open up the original question.

Andri Sadlak 00:41:37  And here the a brand are loud and encouraged by Amazon to answer these questions better. I don’t know if many people have seen this, have tried this. This used to be purely Q&A for customers. Now it’s replaced by ruthless. And that’s how we get to this stage. Now, obviously there are other ways you can do this, and some people claim this is great hat, but you could ask other shoppers to ask more questions and provide answers again, right? I just have to be very careful and not to misuse this and not to break Amazon toes. So just use this at your own discretion. But it is another way. Ruthless is validating the validity, the correctness of the answers, right? The actual user inputs, the Q&A and the reviews reviews. Obviously we can manipulate we shouldn’t, but Q&A, we can answer the questions better and that answer can be used in the average answer coming from Rufus. Does that make.

Josh Hadley 00:42:20  Sense? Would you basically like if you were to optimize this, would you almost create like a page long like response to somebody’s very simple question that it’s like stacked full of like other keywords and other like phrases and answering maybe even other questions in that answer in question.

Josh Hadley 00:42:37  Like, have you tested that? What are your thoughts on that?

Andri Sadlak 00:42:40  Yeah. Funny, they asked that. We did test a lot of different things, sometimes stupid things too. Amazon tends to remove responses that seem to be inappropriate, and it’s hard to understand how they classify inappropriate. But you don’t want to go over the top. My go to approach is we answer the question directly the way we know is correct, factually correct. It’s going to help the customers. And Amazon is customer centric and we try to blend in there. One of them causes more relations. So we can answer the question in a way that says, yes, it can di you here. And this is specifically built for dive here. And then it explains a little broader picture of the use cases. Right. So I wouldn’t make it too long I wouldn’t stop it with keywords. That’s a big no no. Just answer the question. See if you can enhance it with a little bit more background on what the product is, who it is for, and so on and so forth.

Josh Hadley 00:43:26  Love it. Makes sense.

Andri Sadlak 00:43:27  Okay, so in summary Q&A seating, which is the third lever for optimization, we can add questions to listings. We can submit brand answers to enhance with customer relationships, as I mentioned, and Amazon customer can also answer. So keep in mind if somebody asks a new question. This way, Amazon will email some of the recent customers who bought the product recently and new customers. Sorry, recent customers will come in with new answers too. So something to keep in mind too. And again, please do not weaponize this against your competitors. Do not deceive shoppers. Use this properly just to enhance the information. Our goal is expand the data to win this identity e-commerce game. Okay, so now whatever. Number three, the citations or the trust layer. Again we want to audit identify which external sites Rufus is already citing. We’re going to enhance. So build out our own website with BDP pages with all the specs for the products and publish publish blogs, the answer shoppers questions.

Andri Sadlak 00:44:19  And number three we’re going to seed. So we’re going to approach affiliate media with affiliate commission offers and send them drafts of the actual product reviews so they can publish. Make it easy for them. Let’s dive into number one. So can we actually earn rufous recommendations through media? Simple answer is yes. The way you can test it is ask any question. Rufus. Add at the end with sources and it’s going to show you the actual sources that are being used. So you can click on the sources, see the actual media that Rufus is using externally to validate the recommendation and recommendations it’s given to you. Side note you can also see Shop Direct products so products that are not sold on Amazon. If you sell off Amazon you have big volumes. Drop me a message. I’ll share the email you can reach out to on Amazon to get connected to the Shop Direct program as well. Now back to the sources. There are ten signals that make a page suitable based on our research so far. I’ll go through them quickly.

Andri Sadlak 00:45:07  Number one, the URL slug has to contain the exact query. For example, bastion pool for curly hair. Number two title mirrors that query as well. So bastion poof curly here in the title here. So if it’s the latest 2026 top picks Regency matters, then date data has to be recent. And if it’s updated, last updated date has to be visible on the page as well. Then site has a clear niche. So it has to be relevant, not just random review sites. So if it’s a beauty magazine, for example, specialist blog high likelihood is going to pick it up. Then site has to be indexed and cited elsewhere. So not brand new, then listicle or rank comparison format. It is different for every other MLM, but Rufus for some reason really likes the best products for curly hair top best shampoos, whatever. So these listicles and rank comparison format websites is where you want to be. Then each product gets a distinct blurb. So not just enameling, but actually explaining what this product is about and who it is good for.

Andri Sadlak 00:46:00  Then product description with use cases and attributes like ingredients and benefits. Key things people care about have to be in there and then product link product linked directly to Amazon. So very often you’ll see on those listicle websites they’re using affiliate tags in the link. And Amazon associate disclosure will be visible on the pages as well. This is where you want to be the easiest way. You just see who shows up in those sources. Reach out to them and tell them, hey, here’s the checklist, can you follow it? And we’ll talk about the offer. You can also give them. Right. So in in big picture perspective. This is Rufus station split from the data we collected so far. So about 10.9% of the citations go to brand outcomes, whatever your brand website is. Okay, earning media about 16.4% and affiliate review sites 72%. So simply speaking about brand comm, you want to ensure your site has a product page with specs and you publish how to use or best for blogs. That kind of information that increases your likelihood of being picked up.

Andri Sadlak 00:46:52  Now our focus is going to be on the affiliate review sites, which is the majority of the citations. Okay, and guess what helps with placements in affiliate review sites? Some of you may know, some of you may be playing with this already. It’s creator connections. Okay, I spent some time with Career Connections consulting some brands specifically on that, and I just wanted to ask you out loud, do you use creative connections for your brand?

Josh Hadley 00:47:10  Josh we definitely do.

Andri Sadlak 00:47:12  You do. Amazing. So you’re on top of it. So tell me your experience, but I’ll share some of the stuff I’ve been working on previously. So here’s an example. Brand number one. 5 million of extra revenue came from creative actions in year one of doing this program, and we spent 27.6% on commissions. So if you compare it to Amazon Acres. Pretty decent cost to run this program. And they got 12% of the sales that year from Amazon Career Connections. Another quick example, brand number two, 2.4 million and 32% of that was in commissions.

Andri Sadlak 00:47:42  And in that case, 36% of the sales came through career connections. Okay. The big question is what do you think these brands have in common? And it is.

Josh Hadley 00:47:51  I mean, these guys are all like using really probably good affiliate commissions, and they’re probably being super aggressive with like how frequently they’re posting these.

Andri Sadlak 00:48:01  Pretty aggressive. Yeah, I was running this for them. And both in the beauty space just happened to be. And it is one of the spaces that is good for career connections. But it doesn’t mean that you can do this for any other niche. And the real common thing with those is the majority of the content that actually led to sales is article based. So a lot of people think about influencers. Just people are recording Instagram stories or TikTok videos, but articles actually brought most of the sales and some sometimes big names like wired or the New York Times and so on and so forth. You don’t have to be in those, but you can still leverage career connections for Rufus. And this is the framework you can follow so that not only you get extra revenue to create connections, but it can also show up much more often in Rufus.

Andri Sadlak 00:48:38  So you can send the draft copy to the media that you see is being shown in sources by Rufus to make it super effortless to place you as the number one topic for that problem you will ask about follow the ten signals that make a page. Rufus that I shared before. Number two offer this information. I would say 25% plus usually gets you into the door and they’re like, yeah, if you’re going to be number one, I’m going to make more money. I see you a good seller. Why not? Let’s test it. So you kind of want to make it an easy offer to say yes to. And then you want to build a relationship. So check in quarterly, make sure your articles are updated. And again Rufus is prioritizing the most recent articles. So you want to be ranked first all the time right. And the goal is to own what others say about you and influence Rufus this way. Who’s excited? Hopefully everybody decided.

Josh Hadley 00:49:22  So Andri, how do you give like people in creator connections like the draft for that? Are you literally just like copying and pasting that inside of like the The Creator Connection campaign to get the questions?

Andri Sadlak 00:49:34  Yeah.

Andri Sadlak 00:49:34  This is yeah, more of a technical question to to the workflow build. Basically the way we do it is people apply to join the Brand Ambassador program. The way they apply is there’s a link in the description, and there’s a link in our answers to their messages in Career Connections to fill out a form to join our Brand Ambassador program. And in there they leave their email to answer some questions and we can follow up with them saying, when you create content, here’s what we need you to do. Obviously we can ask much more. Can you post on YouTube? Can you post a blog post article on your website? So there’s a little bit of manual process that can also be automated by agents by now, but that’s how we communicate with them. We don’t just leave them to do whatever the heck they want, you know?

Josh Hadley 00:50:12  Love it. That makes a lot of sense.

Andri Sadlak 00:50:13  Okay, so I’ll summarize the two lovers just so we have a full picture. Lover number one is Cosmo the knowledge layer.

Andri Sadlak 00:50:20  We optimize the browse nodes. Item integrity. Backend taxonomies are critically important to have back end all answered, and we address all 15 relation types, all of them that apply to you. Then Rufus library number two the conversation layer. We want to collect the questions that people tend to ask in the category. We want to test those answers on our listing. Then we want to optimize copy images and A+ to answer these questions as well as possible. And number three, we want to see the Q and A’s in the appropriate section of the listing to make the answers even more complete. And level number three citations the trust layer. The outside of Amazon layer. We want to identify which size Rufus is sighting. Build out our website to address all the possible questions people may have. Write blogs, have PDP’s on our website as well. And number three, reach out to affiliate media with Creative Connections Commission offer so that we become number one best suggested product. In simple words, if you only have time for one thing to do, focus on lever number one, which is the Cosmo lever that allows you for us to know your product.

Andri Sadlak 00:51:18  Everything else is secondary. If you without that knowledge layer basically leverage two and three just have nothing to build on. Okay. Very important to understand that cosmos, that fundamental layer for AI. And if you want to scan the QR code, you’ll get the cloud skills that I mentioned. Basically files you can upload to your own cloud account and just do everything we discussed practically for your brands today.

Josh Hadley 00:51:39  Andri, this was a masterclass of AI optimization for Rufous on Amazon. I genuinely have not seen anything better, more comprehensive than what you just showed today. And I think like this. I’m sure there’s going to be a part two that will probably have to make some updates to here in another six months, as the world of AI continues to evolve. But Andri, I love to leave the audience with three actionable takeaways from every episode. Here are the three actionable takeaways that I noted. Number one, you need to understand that there are two now algorithms basically running your search visibility on Amazon. There is the Rufus and Cosmo visibility function.

Josh Hadley 00:52:17  And then there is kind of like the legacy search, keyword optimized version. And those brands that do not begin running both of these systems and optimizations simultaneously are going to be left behind. So that’s action item number one. Action item number two is I would begin optimizing your products today. And I would start with like the lowest hanging fruit is literally sitting in your existing product listings. Yes we could be doing creator connections. Yes you could be reaching out to PR agencies. Those things are good. But the easiest thing that you can go do today is update your images. You can go update the back end information of your listings. You can go update the alt text in your images. Those are all things that you have control over today to help take action on this. And third, last but not least, this may seem very overwhelming that there’s like an endless number of things that you could be focusing on with your keywords. But what I would do is action this down to like one step at a time.

Josh Hadley 00:53:16  So I’m going to put you on the spot. Andri. And say for action item number three, out of everything you just covered, if brand owners could say, man, there’s like 20 things I need to go do. What is the one thing that you think is the biggest needle mover out of all of the things that you listed?

Andri Sadlak 00:53:30  Say the absolute main focus should go into optimizing for cosmos and making sure the basics about the product all answered in the back end and in the front end. That’s the thing you need to be doing anyways. And everything else, as I said, is building on top of it. So optimizing for those questions is the second most important thing you can do. Optimize the citations. So making sure you show up outside of Amazon is a very much a long term play that will help you solidify your positions. Yeah.

Josh Hadley 00:53:55  So there you go. First things first. Make sure you go upload that to that bulk file and that flat file for your listing. Make sure everything’s filled out properly.

Josh Hadley 00:54:03  So that’s a big action item. Love it. Andri final questions here. We’ll do our speed round. Number one. What’s been the most influential book that you’ve read and why?

Andri Sadlak 00:54:11  I’d say it’s, in terms of brand building, I’d say it’s the building a story brand. This one is not really about the AI optimization, but more. How do we build a brand that people want to stay behind, stand behind, and support? One of my favorites building a story brand. Look it up if you haven’t read it already. Definitely worth a read. And if you are into brand building, you’ll really, really enjoy it.

Josh Hadley 00:54:30  Fantastic recommendation. Question number two what’s your favorite AI tool that you’ve been using and how have you been using it?

Andri Sadlak 00:54:36  I might be biased because I helped build this OMA, but as always, my favorite tool in a genetic commerce optimization. And if I wanted to say another one, I probably spend the most time playing with Claude. Lately, Claude has incredible tools, including cloud code work and cloud code.

Andri Sadlak 00:54:50  So if you haven’t already tried, you’ll be surprised how much easier it becomes to build now using these tools.

Josh Hadley 00:54:56  Andri, thanks again for your time today and joining the show.

MC 00:54:58  Thank you for listening. Visit Ecom Breakthrough Comm for more information. If you’ve enjoyed today’s episode, the best way you can show your appreciation is by clicking the subscribe button and quickly leaving a review. See you again next time!