Amazon's Secret Algorithm Shift (Don't Get Left Behind)

I think we already seen from last one to two years that the discovery of what to buy is being done by AI. Is your product good or not is going to be not decided by humans. It's going to be decided by AI and the checkout will still be the Amazon game. >> As a seller, what do you need to be doing right now when it comes to AI? >> AI is as good as the input and the context. If you give a wrong context, if you just say, "Okay, here is my Amazon." And just tell it to optimize, it's going to go in all different directions and you'll end up in actually making it worse because you did not give the right input and you did not know what are the right questions to ask. >> You know, there's a lot of tools out there that are AI. What frustrates you out there that you see that people are getting I don't know if conned is the right word, but maybe so conned by >> scam. >> It's not AI. You're watching Marketing Misfits with Norm Ferrar and Kevin K. >> Mr. Ferrar. Another day, another episode. This is going to be an exciting one. >> It's going to be an exciting one. Our number two of the day, by the way. >> That's right. A lot of you don't realize we record these. We record them on Tuesdays and they do come out on Tuesdays, but we record about every two to three weeks and we typically do three in a day. So, we try to get ahead of it. Uh, so sometimes the episode comes out within a week or two. Sometimes, uh, >> it's been months. >> It's been months. And sometimes we forget about somebody and they they message us and like, "Is that episode ever going to come out?" We're like, "Oh, yeah. I guess we missed that one. Let's go." And we go to the editor and we're like, "Where'd it go?" And they find it and uh and then it comes out. >> That has happened. >> That has happened after this for a couple years. Uh, that that's what happens. And for those of you who don't know, we also have a newsletter now, misfits.news. So if you go to misfits with an s.news, you can actually check out our weekly newsletter. It comes out every Wednesday. It's totally free. And we we what we've done is since we've been doing this for a couple years, we've taken the episodes and we've organized them by topic and we take typically two episodes that are on a similar topic and we put it together in one newsletter and we summarize some of the key points, the action stuff, the software that was mentioned. We add a couple new uh cool little sections to it. One's uh you know the humidor from Nor and I. It's got some interesting stuff on whiskey and cigars. The reaction we've getting, we just started it about a month and a half, two months ago. And the reaction we're getting from the, you know, it's growing. It's small right now, but it's growing. The people that love it that are getting it. It's very concise and easy to read. And so I'd encourage everybody if you're a fan of the show to actually check out misfits.news news uh and uh subscribe uh and plus you know when you subscribe we actually send you uh a PDF of nine of the uh top strategies that have been shared on the podcast from the last 2 years and then if you refer two people after you subscribe we give you 27 more of the top stuff. So there's some valuable stuff in there that can really help you in your business. So we normally would personally like to invite every one of you uh to to check out the newsletter uh at misfits.news. You know, AI is a big topic right now, Norm. We just have we just recorded an episode on AI and we got another one talking about some AI right now. It seems like everybody wants to know AI. Yeah, absolutely. And uh it looks like uh the next episode and the next episode, possibly the next episode will be on AI. So uh yeah, no difference in this one. We're going to >> Yeah. Go ahead. >> Yeah. And this one though, I mean, our buddy that's coming on here, he he was in Nashville with us. >> Yeah, >> we met we met him at the Prosper show. Uh he he comes from a world of uh you know, he was doing a lot of tech at Meta and at Apollo.ai and uh he was at Wayfair and several other places. And then he said, you know what, I see an opportunity here where a lot of people are not using AI, right? Uh you know, there's a lot of people that stick AI on their tools or stick AI on their company name and there ain't no AI behind what they're doing. It's if then else statements. It's conditional stuff. It's just a set of rules. It's not really AI. There's no thinking or going in or real real deep stuff. And he's like, you know what? There's an opportunity here. And we're going to talk to him about what he's done. It's primarily for Amazon sellers, but it can apply uh probably to a lot of uh other people as well. So, it's going to be an interesting talk and it it's kind of the way of the future for the more advanced sellers of what they're doing and how they're getting a competitive edge. So, we're going to learn about that today and it's going to be interesting. >> Yeah. And just before we we bring him on, I saw something yesterday that blew me away. And this was a robot. Okay, just a robot. And it was a figure skater >> skating on ice. >> I thought that was crazy. >> They just had a robot in China win a marathon. It actually beat all the humans in a marathon. Uh running like running like a human, you know, not like scootering along, but like running like a human. Yeah, it's it's getting uh next level. You're going to have a dance partner, Norm, pretty soon, you know, since when >> you're going to replace you >> when you Yeah, it's You can dance a lot better than me. It can hold hold you up and do the turns and everything. Uh so when you want to go twosteping, you know, you're you're going to have uh something that's never going to turn you down. >> Oh, finally. >> Memories of high school. >> Let's bring in our guest today. He is uh Jamon Patel. Jamon's from a car uh which is an AI company that he just recently founded. Uh how you doing Jamon? >> Doing great. How are you Kevin? Nice to get meet you. And how are you know? >> Good. Good. Thanks. >> Yeah. So you're you're in the you've been how long have you been doing this AI stuff? You were doing corporate AI some AI in the corporate world for a while, but how long have you I know most people are like well AI just has been around since Chubby GPT came out, but that's not true. It's been around quite some time. How long have you how have you been how long have you been dabbling in it? >> I say probably a decade. I think uh I started doing it probably in 2016 and Wayfair as well. I did like ML modeling but at that time it's more like ML machine learning. Uh but yeah probably a yeah almost a decade now. >> So you you're doing something called called context intelligence. Uh what the heck is context intelligence? >> Yeah. Yeah. So I think the uh the best way to explain that is uh let's let's look at how humans work. All right. And let's take an example of a uh a seller who is hiring someone like a junior PPC manager. What that person will do um first they're going to check all the data and they'll check okay like okay what are your brand goals and they'll understand how Amazon is working for them. They'll understand their e-commerce tech and gather all the signals. sit down with the brand owner and understand what matters, what has worked in the past. uh so there's a lot that goes into human brain and I think why humans end up doing better a lot of the time historically uh is because of the context because humans have the context that often uh the machines are missing and they only think like one or the other way which was the traditional ML world like you you do all the prediction you train the data and they kind of do all the prediction on inventory or or even the bidding but where it really has been failing is the is the context part and the context intelligence is essentially that which is basically mimicking the human behavior and human intelligence. Uh tying that back to how kind of AI can be developed around that and how can AI become close to a junior PPC manager and obviously it will evolve to senior PPC manager after that. But yeah, I guess a good way to start that is probably a junior PPC manager is probably how I would put it. >> So context is like if the lights are so the lights are out in my room. I'm sitting in my bedroom at night and the lights are out. So the computer knows lights are either on or they're off. Out or on or off, right? That's basically the computer. It's a a state. Uh but context is like, well, I know the lights are off and I know because I can see through my window that the street lights are also off. So maybe the power is off. So that would be that would be a context that I could pick up as a human because of the other environmental stuff around >> that a computer could not pick up. Is that >> this is this is when the sexy music comes on. >> That's when the sexy music comes on and I call Norm. >> Now we have all the context. It's like okay forget machines like norm anyway. U so so so how so when you're applying that so you you built this company a car to come in and it's kind of Norm and I actually uh toyed with this idea about a year and a half ago cuz we were we started Dragonfish a couple years ago and we uh we have parted from our original weeks of work of what we were going to do on it where it's going to be managed services and stuff and one of the things that we were looking at is like after like why are we hiring all these people AI is taken. I was like, why don't we actually use MCP? This is an up and cominging thing. And why don't we actually tie in all these agents that they could talk to each other? The PPC one can talk to the the listing optimization one and the listing optimization one can talk to the product research one and they can have this like chat and we have a human in the loop that's kind of overseen and that was my idea. But we decided uh you know that looks good, but that's going to be hard. Uh that's going to require more money than we had to develop that and some really smart people to actually do that. So we went down, you know, now we're doing AEO and email stuff. Uh, but is that basically what you're doing? So you're basically you're creating a brain for an Amazon seller to where it actually talks to all the different components from listing optimization to inventory to PPC to uh customer service to whatever it may be. Uh, and they kind of are optimizing on each other and fixing stuff in the background kind of automatically and then a human steps in when they need to. Is that is that a fair summary? Yep, that's uh completely right. Yes, I think um and and you're also right about you know you know it's not it's not that straightforward right? You know it's not like giving a prompt and then let's just AI figures it out. We all know it how much it hallucinate and how much of the guardrails and the understanding AI needs to have. Uh but yes that's the idea here and that's what we have been building which is as is combining all the different data points and uh having that context is what kind of serve as an optimization layer uh to kind of do your PPC do your listing optimization or anything else that going to be expanding on Amazon and beyond. >> So how much human involvement has uh does there have to be? >> Yeah. So I think I think there are two parts of it here right like so first building the initial context there needs to be some input around your goals you know your campaign level structure like how do you want uh your campaigns to go because some sellers would like to go aggressive or some are like this is my guardrail do not ever you know go below or above these tacos and there are a lot of those kind of guardrails and brand context that needs to be in place uh because I do believe that you know working at uh working at Meta Apollo and other companies like context is or or let me put it the other way AI is as good as the input and the context right if you give a wrong context if you just say okay here is my Amazon and just tell it to optimize it's going to do a lot it's going to go in all different directions and you'll end up in actually making it worse because you did not give the right input and you did not know what are the right questions to ask so the human in the loop is is kind of there is stage one which is kind of helping AI to build a context with your goals and then the stage two is more more around kind of reviewing and transparency. Now some of the sellers would decide to kind of put it on autopilot from day one while others would review and uh approve. So there is a human in the loop. There is full transparency. So they can see what AI is doing, how it's thinking and what context it's using. And if it's not really meeting your expectation, you can give feedback and that is kind of the other another human in the loop layer. And that feedback goes back and kind of improve the next decisions that AI is going to take. >> Hey Norm, do you know any sellers out there that are just burned out doing this uh ecom game? You know, I I know a lot of people that have talked to us, you know, when we go to events, and it's not only that, they don't know where to start. And >> who would you recommend they talk to? The >> first one that comes to mind is is Quiet Light Brokerage. And here's why. They're going to build you up. They're going to understand your company. And at the end of the day, you're going to know how to maximize your valuation. So, the very first thing you need to do is go and get your free confidential uh valuation at quietite.com. They're going to ask a couple questions. Uh you're going to meet up. It's one- on-one with uh somebody over there and then, you know, let the games begin. Awesome. What was that website again? >> It's quiet.com. Awesome. I'm going to head over there. Now, I I want to talk a little bit about the human in the loop. And this is kind of a carryover from one of the questions we had in our last podcast and that was getting humans to take part and answer the questions. Do you find that uh you have a lot of uh time between getting the answers and getting humans to actually give you the input so that you can start the operations. So you mean kind of uh is there >> on boarding process he's saying is how difficult is it to get this information that you really need? >> It's not even the onboarding process is whenever you ask. >> Yeah. >> Oh, I see. Yeah. So I think it it it's like 80/20 I would say like 80% of the data uh we could really feed into the AI with context using the APIs as well as external signals. Right? So there is so much context that can already be built with the historical data. Now the 20% is what kind of sometimes obviously it's hard because you kind of need to make sure that you get the right input and a lot of the sellers will kind of miss it or not really understand that okay like here is where I need to give the context here is like my campaign structure itself is wrong to start with and then you put AI on top of it to optimize it you're not going to get really amazing results because your foundation is not really right. Uh so that's definitely definitely challenging. Um I think it's probably a way of your user experience and how you build it in the app. Um and that is why we come in as well like you sometimes want to meet with sellers in one-on-one and like try to kind of answer those questions because a lot of the time like fully self has their own kind of u disadvantages. >> So walk me through this. So if I'm if I'm uh if I I meet you at uh Ecom Mastery AI, I talk to your booth like, "Oh, this sounds pretty cool. Uh I want I want to try you guys out. What do I do? What happens next? What's the Walk me through the steps of what happens next? >> Um if you would. >> Yeah. So um the process would be you know just signing up uh connecting your APIs and uh there are a couple of um features they would be giving your >> in my APIs and my Amazon API. >> Yep. Amazon APIs. Right. So the process would be that you kind of sign up uh you connect your APIs. is like an onboarding checklist uh that you would go through. Um and we already have that kind of pretty laid out in your uh dashboard. You kind of do step by step. You you give your input in terms of your goals. Uh and then we would meet one-on-one because we want to make sure that you are set up for success uh in the initial period when we kind of uh review the account with you uh and set up the AI in the right way. So we would like activate and after that once the data is synced AI will already analyze lot of the things like it will audit your account. It will find the opportunities. It will uh look at the campaigns. So it will have all the context by the time we meet and then during like during live meeting if you have any questions we would activate agents in your campaigns. We could also launch new strategies. Uh we would look at uh AI to kind of optimize your listings, images, title bullets, keyword research, all that will be done. Uh and then from there it's mostly autopilot. You can review it once a week or if you decide to manually approve that might you have to probably log in every day and manually approve every optimization recommendations done by >> was it making the changes for me or it's making the suggestions I have to go do it. >> So that is a that is a a flexibility kind of that's up to you right so as a seller you can decide hey I want to automate it completely and it's not going to wait for your approval it will start going into your Amazon account and start implementing it. Can it give uh a couple of options and then you go and approve and then it automatically posts it? >> Yes. So exactly. So the other other approach would be manual approval, right? So yeah, you couldn't like just approve and then once it's approved it will implement it. >> So is this PPC only? You keep referring to PPC or is it actually doing other things besides PPC? >> So it does uh listing optimization as well, keyword research as well. Uh so we kind of do similar uh way where it would bring all the context and the keyword research and then pass that to your uh listing optimization including images, title, bullets. So it has understanding of your uh keywords, your listings, your brand, um rufus, what has worked in the past and and based on that it optimizes listing. uh it also uh optimizes kind of your inventory and we are getting into more and more features but right now uh it's mostly the major two areas that that we are focused on is the listing optimization and PPC as well as the MCP which is a different uh topic. Yeah. >> So is it creating the images in my listing that's actually create so you're using you're tying into one of the other outside models that's creating that underneath the hood. >> Yes. >> And is it strictly your slide deck or do you get into the uh A+ images as well? Oh, A+ as well. Uh that's not released yet. It's a more of an internal launch right now. But yes, it will the idea is that it it should be able to with the innovation in AI. Uh not only A+ videos, uh creatives for advertisement uh as well as you know beyond Amazon, you kind of need creatives to for meta ads, Google ads, Tik Tok. Um, a lot of sellers as you all know like miss videos because it's high effort has been high effort in the past uh to put videos in your listing and that should also be uh possible to do. Right now it's limited to images. >> Now what what about affordability? Can a smaller seller afford this or is it more towards the intermediate? Uh no it's uh I mean we often see that I think smaller sellers should be able to afford I think a lot of the examples we have seen sometimes is that u even the cost for these kind of platforms is the AI cost and the compute cost as well as the database kind of you manage the data right smallers will have a less number of campaigns less number of products uh so it's also not kind of expensive for platforms like us to also manage them so it's it's based on the seller size anyway. So it is affordable. >> So you you you have an MCP um so other AI systems can talk to your system, right? So if I want if I if I'm running Google Workspace or I'm working running Slack or Teamwork or one of those things, I can actually tie that in uh and say if I got a PPC manager and say, "Hey, uh this campaign's out of whack." it'll ping that ping them and say, "Hey, you need to go check on this or something like that, right?" Or I can tie in data, different data points into your system and and supplement it. >> Yeah, exactly. So, I think MCP is the next topic I'm like really really excited about. Uh, and it's actually shifting a lot in the industry in terms of how sellers work. Uh so just to give you an idea like there there were um I think around like 10,000 like 100,000 uh toolkits that developers using for MCPs and that has grown to 97 million in 18 months and uh it's crazy adoption of MCPS in not just on e-commerce but everywhere. So um >> explain what MCP is for the people listening that might not know what that is. >> Right. So uh think of MCP as your uh connector between the AI assistant to your system whether it's Amazon uh take an example of Google calendar or even a slack or simple like Google email. Now your system is Google mail but that system needs to talk to claude or open AI the way they interact with each other is through MCP. It's a model context protocol built by claude anthropic and it's basically a way for AI assistants to not only interact around the data, it can also act on a data. So it's more of as an integration. However, in traditional world, if you look at how things were being done is you know you would have a developer building APIs and you'll build the integrations one by one and like you'll it will increase your effort so much. MCP basically cut downs all of that uh where you just have a central layer talking between AI and your system. >> So like API is like USB 1 USB 1.0 and uh MCP is USB 2.0 where the data goes both ways to one way. Okay. >> Yes. Yes. I think and the USB 3.0 is still built on USB 1.0 because you're still have the API you still have the API that needs one is HDMI and one is uh uh USB 1.0. Yes, but it's a it's a it's a right way to put it. Yes, it's definitely an upgrade. Um and I think that that has been unlocking huge amount of possibilities, right? So I let's talk about Nashville like I talked to a couple of sellers and most of them were really doing the workflow where you would have a cloud co-work and it will download Amazon reports right so you put it on your computer folder and every day it just downloads much of reports your brand analytics report system report and then your AI is going in pulling those reports analyzing that but that's like super limited manual it's just report by report now think what IMCP unlocks with cloud. Uh a lot of people actually don't know this. Claude launched live artifacts. What does that mean? Is that you can literally build like a live application that can refresh every single day. We are not talking about static dashboards. We are talking about custom applications that can be built uh through any of these MCPS including us. Um and you can use these MCPS to build anything you want. And I I was talking to a seller and then we built just like a custom application u local to his computer pulls all the insights about his accounts um build the charts the way he like to see it in within just 10 minutes and that's the power of MCP. >> That's the power of now the data like you're you're overlaying on top of the Amazon data but now with MCP I can create my I know guys are using artifacts for like stocks and stuff. They're creating their own stock uh ticker and everything. you can go in instead of having to rely on, you know, one of these off-the-shelf tools that says, "Here's your dashboard." And you're like, "No, but I don't need that. I need this. I need this, and I like to see this first." And you basically just talk to it. You can almost viode this. You don't need to like be a coder. And you say, "This is what I want to see. This is what I want the report to be. I want the this uh if it's I want to be red if it's a problem and green if it's uh if it's up or whatever." And just does it. And so you get it customized to exactly the way you like to work and what you like to see and what's important to you on your KPIs. And that's that's the advantage. Um, but I but but why do I need your tool? I could just use MCP and go into I could use Claude and and go in and pull these reports and create my own dashboard and tell Claude code or cloud co-work or cla code, hey, okay, go execute these changes uh based on this set. So why do I need why do I need a car? >> Yeah, great question. So there are a couple of uh areas to think about here. First is let me start with security. Uh if you're just using like random API and trying to kind of put together an MCP uh which may not have all the security uh features built in, I think that's one reason you would want to uh make sure that the MCP built by a platform is reliable enough. The second uh is the raw API like there's a difference between you kind of call a huge database and tell cloud to just call the API or the huge database table and then trying to pull data you're going to run out of tokens. So the MCPS built by us are going to be really designed specifically for Amazon seller use cases. Um there are tools that are designed uh the guidance is given. Uh there is so much accuracy and iterations that are done to make sure that the clot doesn't hallucinate. Uh so if you just pass the table like a huge database table and tell clot to really analyze, not only it will go out of tokens very quickly, it will also elucinate and try to really put together something that is not probably accurate. All right, so that's second. And then the third thing is intelligence. Um a lot of the time if you just want to analyze the data yes you can probably let's say you can plug into any MCP that is really reliable enough um and you can just plug into that and you get the data. There's a difference between the data and intelligence. So with our MCP which is the next phase we are working on is the intelligence that we have been building is going to be tied with that MCP. So what really happens is that all the context that we talked about and all the intelligence that we've been building on every single feature is going to be part of the MCP part of the tools skills and the guidance that is coming with that with that and so that way you don't have to like start from scratch and just trying to figure out what are the right questions to ask how do I tell cloud to do what um and it it can just go into any direction resulting in any kind of uh recommend condition that may not make sense at all. Uh and then the fourth thing is actions. Uh it's actually very risky to have uh an MCP without really reliable and and an end to end platform to take actions because it can wipe out your entire Amazon if you don't if you don't do it right. Right. It can just go into if you give an action or edit permissions that's a whole different story and that is the next phase as well. uh we have kept it read only for now uh but that's something we are looking into next which is how do we really make it reliable to take actions uh from claude >> so we talked about that again on this last podcast uh the circumstances around security people a lot of people don't understand it that endto-end point security is it's critical but the other thing I just want to relate this to uh the consulting business so a lot of the times you know we'll we'll have conversations with people uh about whether I should go into an agency or not. And they don't realize that they might give a little bit to an agency, but they're going to grow their business tfold. They're not going to pay the Amazon tax with all the mistakes they're going to make. And this is where you pay an expert. And that's the same thing with you. Maybe not as an Amazon consultant, but I'm looking at any um any anybody who's involved with AI, if they're creating a platform like what you're doing, you know what you're doing. And if I just come on and I'm just going, "Oh, uh here's my chat GPT free account and I start playing around with it, I can mess things up or I'll get just zero performance out of it." And a lot of people instead of paying a little bit extra would rather just screw around and not prompt properly, have poor security, and at the end of the day just not get anywhere. >> Yep, that's absolutely right. Yeah. >> So, you guys are basically you're cleaning you're because you're a seller yourself in the past or still selling, you actually know what sellers want. So, you're applying those principles and that context to it. you're actually pulling the data and you're organizing it, you're cleaning it, you're securing it, and then you're opening up an MCP like now, okay, now if you want to go customize it instead of us building inside of our platform and having to do all this uh advanced code and everything, go use the power of claude or whatever your favorite tool is and actually layer on top of that and just just just go to town on whatever you want to know about your data and about optimizing your business. So, so that's that makes sense and I think that's where a lot of of stuff is going. We've Norm and I have talked about this where tools like Helium 10 and those guys I think they're having trouble getting more people into the tool to use it because the dashboard is like this is the dashboard. Oh yeah, you got some suggestions for the dashboard, you know, hit this button and send it to the dev team and maybe we'll add uh you know, this feature versus now you don't need that. You can just go to claw and say I want this, this, and this. just talk to it like a human >> and it it'll it'll spin it up. And so someone like Helium 10, their value prop now is not going to be in their software suite. It's going to be in their data because they've been scraping Amazon since 2015 or whatever. And Claude doesn't have that. Uh I'm assuming or it's not nice and tidy. Then you can come in over layer on the top of that clean it in the way that needs to be cleaned and then you put on the put clo or whatever or co-work on top of it and and really dive deep and ask it questions that you could not otherwise do and get insights and find the the gaps that would be very difficult to do otherwise. >> Yeah. >> Yep. Exactly. So, >> so what what are you seeing like in the you know there's a lot of tools out there that are AI uh tools and AI for those listening I'm putting air quotes what what what frustrates you like when you when you go to a van you go walking by the booth you're like oh god that's not freaking AI that that's just you know what what what frustrates you out there that you see that people are getting I don't know if conned is the right word but maybe so conned by uh where they I think they're getting >> scammed scammed by >> and it's it's not it's not AI. Um what what what frustrates you? >> Yeah, I mean having spent like years in this industry and building AI products and ML products, I would say I think when this whole shift started happening in probably like few years ago. Um I think one of the area which is I would call chat agent. Uh you know you talk to you you just type something in the chat and it answers something. A lot of the platforms I I've noticed you know that without having a single like intelligence feature you just you know wrap around it as a chat uh just answers basic questions very very basic questions and that too not at a level where you know nowadays claude or opening I can do in a much much intelligent way um that kind of little bit I would say it's frustrating because you're basically you know there are there are companies who have launched like let's say reporting agent but that basically is nothing there's not really much of an intelligence I would say so yeah it is AI but it's not really any intelligence in it uh the second one would be like rules systems you know EPL's um I don't know just calling that as an AI or a putting a small like an ML u machine learning like model and calling it like more of an AI so there are these these things I mean there are probably companies or tools which are not doing any of these and calling it AI uh which is even worse. So, so I think that's another frustrating level that >> what is int what is intelligence? What does that mean? What's intelligence from an AI machine world? What does that mean? >> Yeah. So, I think for us which which is the topic we have been discussing. It's a context context intelligence like that to me is what be like like one of the intelligence uh I would I would call it in the in the AI space. I think there is another one like let's say agent to agent collaboration or communication like anything around agentic as well could be another innovation that we called out. Uh I think there are two levers here like you know how much autonomous they are there is agents that can just perform tasks. So that's basically a workflow that is still like AI but it's more around performing task like you know analyze keywords um put them in a file send an email like that's a workflow. Uh so that's kind of a layer maybe two or something and then the layer three would be agent having the thinking. So to me the innovation or the technology is when the thinking capabilities comes whether it's through context intelligence or something else and that is kind of why the agent world started floating around in the industry because agents can think and reason uh and then the true AI can reason. uh it's it's has some ability to really say okay you know what >> is it though is it reasoning though or is it recognizing complex patterns isn't that what intelligence is the ability to recognize complex patterns and put them together >> yeah I think that's including that as well as like reasoning around you know just thinking okay like I'm a I'm I'm recognizing a complex pattern but then okay what do I do about that right so do I you know increase bit do I not do something do I >> that's probabilities that's probabilities historical. Yeah. >> Yeah. >> Okay. >> So, yeah, I think that's what uh we have been seeing and that's my point of view there. >> Hey, Norm, you'll love this, man. I talked to a seller the other day doing 50k a month. But when I asked them what their actual profit was, they just kind of stared at me. >> Are you serious? That's kind of like driving blindfolded. >> Exactly, man. I told them, "You got to check out Sellerboard. This cool profit tool that's built just for Amazon sellers. It tracks everything like fees, PPC, refunds, promos, even changing cogs during using FIFO." >> Aha. But does it do FBM shipping costs, too? >> Sure does. That way you can keep your quarter 4 chaos totally under control and know your numbers because not only does it do that, but it automates your PPC bids. It forecasts inventory. It sends review requests and even helps you get reimbursements from Amazon. >> Now that's like having a CFO in your back pocket. >> You know what? It's just $15 a month, but you got to go to sellerboard.com/misfits. sellerboard.com/misfits. And if you do that, they'll even throw in a free two-month trial. >> So you want me to say go to sellerboard.commisfits and get your number straight before your accountant loses it? >> Exactly. All right. Yeah. You know, I I remember you and I were at this uh conference, Kev, and uh some guy was trying to get our business and he came up and he was telling about how he was, you know, doing this uh reporting and workflow for these companies. And so he told us about it and it wasn't very complicated, very very uncomplicated. and he was charging $5,000 a month for monitoring and like we we questioned him about it and he like he was kind of laughing because you know people were paying that and he wanted us to do this and we just thought no we're not going to get involved with this this is just a scam you know people taking advantage of people cuz because they don't know you know and hopefully what happens is they did get caught up and then they start questioning this guy But he's not the only guy doing it. There's so many people out there taking advantage of these companies. Like I don't mind making a profit. But when you screw people over like that for something so simple. Yeah. I met a I met a seller at Nashville um and they they called someone to build a claude co-work uh you know end to end claude coowork uh setup and uh initially it was like I think this is the seller story I'm sharing um I don't remember her name or anything but like this is just a story that I'm sharing which is so they called them and like they did all the setup and then suddenly like the pricing came out I was like this making no sense like this is probably like hiring I can hire like three people at this cost and why would I pay you for that like simple setup that you're done for one day. >> So what what what do you think sellers need to know when it comes to AI? You know there's a lot of 2 million sellers or something selling on the Amazon platform in the US that are active right now. What a lot of them they keep hearing this stuff about AI. They're seeing stuff on LinkedIn. They're they're hearing ads and getting emails AI this, AI that. And then some of them are diving in. I know like Rob Green, he's got 20 Claudebots, you know, 20 Mac minis all running and he's got rid of a bunch of his Filipino uh VAS and he's he's like really deeply geeked into it and doing the stuff. Then you have other people that are just using Chad GBT to help them analyze a listing or maybe they're using one of the tools. But as a seller, what do you need to be doing right now when it comes to AI? Do you need to be following every little thing? Do you need to pick a couple things and just go down those holes? Do you need to just sit back and let it all the dust settle and then figure out what you're going to do? What where should you be when it comes to AI as a as a someone that's wants to grow and preser preserve and grow their business on Amazon? Yeah, a great question and I think I'm going to answer this from the angle of me being a seller and where I kind of have experienced because when I started few years ago obviously there wasn't an AI so I kind of started from those manual world of you know analyzing keywords in spreadsheets and analyzing everything like manually task by task. uh so my advice like for sellers right now in kind of this AI shift is happening uh every other week there is something new first I'll probably talk about the manual processes so you as a seller you should first identify where you are spending time doing things manually and sometimes it's not the answer is not like the PPC tool or anything like that the answer is more around let's first make yourself effective Right? Because like like we talked to a lot of sellers and they obviously have like midsize sellers would have like a you know two to eight people in their team two to six people and their workflow of how they communicate how they do their weekly reporting how they you know send emails or talk to new sellers or new vendors like there's so much inefficiencies in your manual processes. So my first advice of sellers would be identify those and some of those could easily be done through cloud co-work I would say and you can literally do it yourself uh without a help of any any platform because that's just your own personal assistant. You don't need that assistant to be really really powerful to take actions. I'm just talking about like automating some of your manual processes. The second thing would be um again like being worked in like this industry of uh tech products for 10 years. I think the data analysis you as a seller need to become a data analyst and I think the top agencies and the platforms who are really good at what they do is because they know what insights they need to get out of the data and I kind of met a lot of sellers they would have like an open claw or cloud running to analyze the data but I think a lot of them are doing it in an inefficient way it will just relying on AI to really pull the data and listening to it blindly is a is a mistake you need to really understand the data uh you know where the AMC audiences are like how pricing impacts the competition and how can like do you reduce the price or do you increase their PPC so there's a lot that goes into the data so I think the the other approach would be u you know how can you really build that kind of a data analyst agent for you uh that can really constantly analyze the data whether it's your team member or you as a seller or you go to any if you're evaluating any platform I think you need to see how much of the data they they care about and they're able to pull insights from. So I think that's going to be the the second and I think the third would be when you're really talking to let's say agency providers or you're doing it yourself. The third phase is your implementation, right? So you have kind of optimized your workflows, you optimized like analyzed the data, then you know all the insights. Now the next game is you know how do you implement that and the implementation is usually the most riskiest uh because you can't just you know have an AI. I've not met a single seller yet who is using open claw to implement uh right. So I think a lot of people have been using open cloud to say okay I'm going to analyze everything but I when I ask them okay what do you do to implement but they're probably using an agency or a tool to really implement everything. So there's still a huge gap between how people are using AI to analyze versus how they're using to implement and for implementation you really need a robust infrastructure whether it's the agency doing it or the tool doing it but you need to make sure that they have a really good implementation in place. I I want to go back to what Kevin initially said about sellers. And I think the seller has to have a good understanding about his strengths and weaknesses. Not all sellers can jump into this. They're either going to be technical, analytical, or creative. And for example, I suck. I hate accounting, but I still went and I learned about accounting so I could understand the health of my business. But I don't implement that. I hire a professional that will do that. And I think you're walking into dangerous territory if you think you can be all jack of one trade getting into this where you don't understand it. So you can be technical or you can be an analyst and I think you have to go out to those people, those specialists and have that done. I mean, you got to stay within what you're doing. If if you're great at growing your brand, then do that. If you're great at, you know, uh I I don't know, brand creativity or packaging or whatever it is, play on your we or your strengths, not on your weaknesses. And I I think a lot of us nowadays, we group everybody into one category, the seller. And you can't do that. You know, the the seller will get screwed over if they try to implement this all at one time. They can learn it. And I think that's a good part. get education, go out to future pedia or whatever these platforms are and understand it, but then guide the specialist to what you want them to do. >> Yeah. Yeah. Totally. And that's where I guess the implementation point also comes in, which is like you you you you don't know what you don't know. you shouldn't be implementing based on just because what AI is telling you on a random open AI chat and it's just it's not going to have any context about your brand. It doesn't know your campaign structure. It doesn't know how Amazon works inside out and like you can't implement a system based on just a simple chat. >> I I have a question about OpenClaw. So, I had a friend the other night that was uh talking to me and he was sending he was saying that he was implementing all his agents through VPS. Now, that was new to me cuz I thought everybody that I heard of was implementing through most for most part Mac minis. But is it safe enough to put it on a VPS and build all the separate agents? uh I mean it's I would say the open claw itself uh is not yet fully reliable right so I think that is why it's it's good that I haven't met a lot of sellers having an implementation system built through open claw and that makes sense and I wouldn't probably advise at this point as well because a minute you because what exactly openflow would do is like they it will build couple of agents as soon as you ask it to do something it will build couple of agents uh dynamically and kind of execute tasks. Now it can do really well in terms of analyzing things and answering questions but if you ask it to really implement then it can go in all different directions and implementing that etc. So even like regardless of any system I would say it's it's little bit risky uh to use open claw for implementation at this point until obviously it's going to get more mature and I think cloud and open open AI are already on it. >> Yeah. So especially with building safeguards into it and uh making sure that the security is involved. >> Where's all this going? What's I mean where do you see where do you where's going to be that pivotal shift? Right now people are still doing stuff the old way. They're still using old traditional agencies still ranking their keywords the old way. Yeah. You have Rufus coming in and it's more intent based and it's more phrases and there there's more data being applied to what who ranks where when. uh especially on Amazon, but a lot of people are just they're just still doing it the old way and they're not paying too much attention to this new stuff. What do you need to be saddling the fence right now and then preparing for 2 years from now, 5 years from now? When do you think there's going to be that that big moment uh where everything just shifts? What's what is it something Amazon has to do or is it just it's just going to be a slow migration for sellers or what do you think is going to happen? It is I mean we all know it is already changing like 250 million users were doing rules based search last year and it's been growing crazy. Amazon has even started recently uh pushing like Rufus as a default behavior based on the query you type in in the Amazon search. So in terms of where the industry is shifting I think we already seen from last one to two years that the discovery of what to buy is being done by AI. So what will happen is that is your product good or not is going to be not decided by humans. It's going to be decided by AI and the checkout will still be the Amazon game. I think Amazon will still be the leader in the checkout which is basically a downfunnel process because people trust Amazon and no one can really beat them on the supply chain, right? So even if Jet GPT or or Gemini or Claude comes up with any more of shopping features, people will still be wanting to check out with Amazon because of the the power of Amazon in terms of that. However, when it comes to discovering products like what should I buy? I think that is already shifting. I would say probably it's 25% rate at this point. It's going to go to probably 50% by next year and 75% a year after. So this whole AEO, GEO, shopping discovery, AI agents helping you what to buy, it's all going to be decided by AI. So what sellers needs to really think about is we are in a very interesting transition period. You cannot go all in for rules because 70 80% of your search is still happening through keywords. So you still need to build the foundation, see what are the other things you need to be doing outside of Amazon for AU and GEO. But then when it comes to Amazon, I think the whole advertising advertisement game will change because I don't think the like there isn't a data available of Rufus to sellers uh that okay like this is how your Rufus is performing and there aren't ads appearing in Rufus. So very soon what will happen is that Amazon is going to probably start thinking about a conversation based advertisement the way OpenAI did and the whole advertisement game will change. Now it's not about the keywords. It's more about how AI agents interpret and how the conversions happen. >> So how do you optimize for that? >> I think it's going to be uh dependent on how what kind of um you know ad structure and what kind of data points Amazon will come up with. I think at this point with the data that we all have, I think you optimize it by a couple of different ways. So first within Amazon, I think focus really really deeply on reviews. and your listing because um we talked about context intelligence. Rufus basically use Cosmo which is basically a context and knowledge graph and that that has a brain in it. So it's basically understands every listing by not just title, not just images, not just one thing. It's building that knowledge graph of every product. So you need to build a product that can be in the top of that knowledge graph for the AI or the Cosmos and Rufus and and you need to really optimize your entire listing end to end. You need to make sure that your reviews talk about it. You have FAQs or reviews also talk about it because it's going to be scraped by this knowledge graph and only those sellers will win. >> So I can do that. I can go I can scrape my reviews. I can look at my competitor's keywords. I can get their backend. And I can do all see what's search query performance report uh product uh opportunity explored all I can get all this data but what I don't have is I don't know about you Jamon I don't know that you have two kids age this and this and that you also have a cat and you also like to travel and go uh you know somewhere uh every to New York every year to visit your relatives or whatever and I I also don't know that you're also into soccer and you watch these kinds of movies, but Amazon knows that and it's all in their database and they're tailoring the search and the AI using that that data. So, how do I optimize for that? Cuz I don't I can't go scrape that out the review. I can guess by taking the reviews and the data that's publicly available to me and I can make my best guesses, but I don't know all that stuff. So, like an AI right now, there's no such thing as ranking. You you you're the answer. And so everybody's answer I can ask my AI the same question you do and we might get two different completely suggestions on what to do based on our history and what it knows about us. So how do I use tools like what you're doing in this cont um u the context uh intelligence to actually optimize for that? Do I need to like use your tool to like figure out what is my avatar? Okay, I sold a thousand units list last month and 820 of them were by these types of people uh that do these types of things and the other 180 I don't really care about. That's great, but I need to find more of theseund these 820. So, how do I what do I do? I mean, how do how does this evolve and how do I need to change my thinking if I'm a seller to actually go with where the puck is going and get ahead of it? >> Yeah. Yeah. And that's a great point on the other side which is the target segment uh who you are selling it to and like every customer profile is different. I think there are two aspects here right so AMC uh has a lot of in-depth data but that is more less of Rufus and more of advertisement right nowadays right so if you go really deep into AMC which is what we are already working on like we we haven't released that feature but like another layer on top of the contextual intelligence is what we are building is it will analyze the AMC audiences it will understand what kind of uh an audience what kind of a target segment is really performing converting for you or based on your product and brand what should what you should be targeting uh for in order for you to grow revenue. So I think there is there is this data that is available to us uh from Amazon and from any external uh data points that is mostly around keyword based system and AMC. Now when it comes to Rufus, Amazon has not made that data available to any of us, right? So the only way some of the other tools which we don't do it is is by scraping. So you kind of go into your Amazon, you would type in 100 different queries, but you cannot really mimic that behavior for every different user. You can only know the answers because you cannot really like create different users and different data sets because that's not going to be a real behavior. uh so I think it's a matter of time when that data becomes available through Amazon and I think that can be a huge gamecher uh and I think that has to happen because otherwise sellers will be going so blind and as a blackbox that okay how do you really know what to do on roofers like who who is searching for why would they make that available that's super valuable that's 30 years of Amazon data they've been collecting because if I go and I ask because you know you can do this you can go and request that your your personal file from Amazon as as a consumer. I've done that. Mine when I did it two years ago was 700 and something pages PDF and it had everything that that's like the key to the kingdom. Why would they give that away or why would they make that public? Why wouldn't they just do like what Meta and Tik Tok has done and say, "Hey, uh don't worry about your PPC. We got you. We got performance max or we got, you know, we know more than you do. Just tell us how much you're willing to spend." >> Uh and we'll just do it for you. And then what does that do to you on on you know because you're optimizing PPC stuff does that overnight put you out of business or what are you doing to actually protect against that happening? >> Yeah. So I think the way we look at is like uh as I mentioned like we already done listing optimization. We are actually looking at ourselves as not really a PPC optimization tool. We are looking at ourselves as more of a context uh intelligence for commerce. And it's not just Amazon. It's for your e-commerce business. And what does that really mean is that we we basically help you as a seller grow and there are tons of things that you need to do even if you have one piece of the puzzle that is automated through Amazon and we all know like Amazon recommendations has never told anyone to reduce beats. So I think there's obviously a trust trust factor, right? Uh but let's say even if that's gets solved and Amazon launch is the best of the best uh AI feature for managing PPC, I think that is still that's just a one piece of the puzzle that when you really look at a midsize seller that is doing five, seven different marketplaces and doing 10 things in their Amazon, there will still be lot of things that need to be done. Uh and what we are really building is the foundation of context intelligence. uh I don't think it will matter if let's say if Amazon comes up with an AI feature we'll still be uh relevant in terms of what we do uh and what we provide through MCP as well. Hey Kevin King and Norm Ferrar here. If you've been enjoying this episode of Marketing Misfits, thanks for listening this far. Continue listening. We got some more valuable stuff coming up. Be sure to hit that subscribe button if you're listening to this on your favorite podcast player or if you're watching this on YouTube or Spotify. Make sure you subscribe to our channel because you don't want to miss a single episode of the Marketing Misfits. Have you subscribed yet, Norm? >> Well, this is an old guy alert. Should I subscribe to my own podcast? >> Yeah, but what if you forget to show up one time? It's just me on here. You're not going to know what I say. >> I'll I'll buy you a beard and you can sit in my chair, too. We'll just You can go back and forth with one another. Yikes. But that being said, don't forget to subscribe, share it. Oh, and if you really like this content, somewhere up there there's a banner. Click on it and you'll go to another episode of the Marketing Misfits. >> Make sure you don't miss a single episode because you don't want to be like Norm. help. >> So, can you share a few real examples of how you affected a seller positively? >> Yeah. Oh, I would love to. So, I think there is a there's one seller uh we have obviously couple of case studies um on our website as well. uh this seller is is uh selling like toels uh it's a black and white brand and the brand name is called black and white or something and that seller was doing around $25,000 a month. It's a small seller. Um onboarded to our platform and they, you know, started optimizing it, put it on autopilot. Uh we just would review with him probably once a month. uh he did $50,000 a month in March and his tacos was like 15% uh when like last year when he wasn't using any of the optimization and now his tacos is going around 8 to 9%. >> Wow. >> And our AI basically made uh like more than 3,000 decisions across his nine campaigns and the new to brand uh users also grew from like 30% to 80%. So like it's one example. There is another seller who is selling uh kind of oils and uh on on that category uh beauty and oil category and they like one of our feature is analyzing search term relevancy. So uh we analyze like 12,000 uh more than 12,000 search terms and negated uh 1200 irrelevant search terms before even getting a click. So our context intelligence can really understand the product you are selling where almost all the tools out there would wait for at least five to 10 clicks. So now imagine uh one search term waiting for 5 to 10 click is probably like $10 at least or $5 to $10. You're basically talking about $15,000 of waste spend for any tool out there who is doing a rule based uh system with saying oh wait for 10 clicks or wait for five clicks and then negate a keyword if I'm getting a zero order. So those are the stuff I'm talking about and that seller uh we also have an inventory system as part of our context. So like there were like 7,000 decisions taken but 4,000 were influenced by inventory where we did not go aggressive on the beats. Uh and that seller grew from uh I think it's like almost like 50% in terms of sales uh and and it's like 75k to probably 110 120k. Interesting. It's cool stuff, man. It's cool stuff. Um, so what's next? >> New feature that you're working on that you're excited about. >> I was going to say we we go to the cigar bar in Nashville. I'm kidding. >> I'm there. Yeah, that's right. Norm >> Nor will be there. Um, we're we're all about the cigars. We even have our own little WhatsApp group uh where we share smokes uh and stuff in. Are you a cigar smoker? >> No, no, I'm not actually. But uh I think I I really uh liked all the events that that you organized in Nashville. And I think you know it's a it's a unique one that I always remember. It's like oh yeah, it's a cigar event. >> I see here on your website though you don't you don't like it too much because it says uh we were at ASGTG and Prospy AI. So I'm like I'm feeling left out here. I'm like, I guess you didn't have a good time. >> No, no, no. It's it's uh it's uh it's something we haven't updated for a month or two, but yes, it was there actually. You should have visited probably like few weeks ago. We had a big banner. Uh but that's that banner was about the dates. Uh but then we removed that >> and and what about the marketing misfits cigar event? I mean, that must have been a highlight >> right now. I'm I'm I'm on I'm changing the whole road map to just building a website with with marketing misfits and cigars. >> There we go. Uh there you go. >> Awesome. Awesome. >> Yes. >> You uh um how do they do that? If they want to learn more about uh um a car or they want to reach out to you or what where do they go and do that? >> Yeah, I think uh we would love to talk to any seller, any agency who's really excited about kind of what we talked and want to see it in action. Uh they can reach out to me on LinkedIn or they can go to akarai.com. It's a lot of a uh it's It's a A k a r ai.com. >> You made it simple. >> What's it called? >> Does it have a meaning? >> Yeah, it does actually. So, it means um it's like shape uh in a in a Hindi language. It's it's basically uh it means shape. It's like a shaping your business, shaping your >> you know life maybe. But yeah, that's that's how we thought about it. >> Okay. Awesome. Well, cool man. I really appreciate you coming on. Norm's got a question for you though before we let you go. >> Right. So, we always like to ask our guests at the end of every podcast and I said guest, I should say misfit at the end of every podcast if they know a misfit. >> Yeah. So, I kind of know a lot of people from the tech side where like you know Meta, Wayfair. uh I cannot think of one single person at top of my mind but I can definitely find someone on the e-commerce e-commerce side who is really shipping features whether it's Meta or Wayfair and and you know recommend them to be on your show. >> Awesome. Perfect. All right. Uh that was fantastic. It was um you know I thought when we were coming on here that you know just talking about Amazon would be kind of off topic a little bit but uh I think we nailed it. Um it was great. It was educational and uh really appreciate you coming on. >> That's what she said. >> Thank you so much. >> You nailed it. >> Yeah. Thank you so much Kevin and Norm. >> All right. Appreciate it. Thanks Damon. Appreciate it man. And uh remember everybody, it's a car. A a k a r. So four four a's in there. A a k a dot. No, five a. Aai. So with ai you have five a. So um >> yeah, it's actually I I couldn't get theai domain, which I'm still, you know. >> Oh, you don't. Yeah, it's it's a it's an akari.com, but I I I'm on lookout for theai one for sure. >> Well, you can just give them 100,000 and they'll give it to you. I mean, that's all you got to do. >> That's all. Sponsored Sponsored by Kevin and Spor Marketing Misfits. That's right. >> There you go. I'll make a I'll make a deal. >> Well, appreciate it, man. It was great chatting with you today. >> Yeah. Thank you. See you. >> All right. Stuff happening out there in the world of e-commerce. I mean just as everybody's got a different angle and different take and to help automation and to help uh analyze the data and tools like a car I think are something everybody needs to be taking a look at if you especially if you're on the Amazon side of things it might be something worth uh exploring and get on the phone and maybe do a demo with them um if that makes sense for you right now but uh it it's it's always there's so many ideas and so many ways to use AI norm like every day we're coming across something different >> yeah But like for our uh Amazon listeners, like why wouldn't you check this out? If it could even make you get 5% extra sales, you know, I'm not sure what the cost is. We're not an affiliate. Uh but yeah, give them a call. Check out uh check out their platform. Speaking of calls, uh if you want you want to give us a call, you can't. But the next best thing is to actually give us a listen. You can actually go to marketingmisfits.co. C. Oh, what are you shaking your head at, Norm? >> If you want to give us a call, >> my segways, you can. But the next best thing is to give us a listen. I mean, if you want to, we're not going to We'll talk to you, but you can't talk to us. Uh, no. Just go to marketingmisfits.co. If you want to talk to Norm, actually, you can talk to Norm. You go to dragon.fish and it says book a call. And so if you actually want to talk to Norm, you go to um Norm handles most of those calls. Um so if you really want to pick his brain and talk to him and see how Dragonfish can help you out, you go to dragon.fish, hit book a call, and most likely, at least right now, you know, we're about to scale up. So this will probably change. But depending on when you're listening to this, um you might actually get to speak with one. But if >> Yeah. But you know what? If you're a pain in the ass, I'll transfer you over to Kevin >> and I don't answer. But uh yeah, so marketingmisfits.co is where you can find out the latest episodes. New episode every single Tuesday there on YouTube, on uh on Spotify, and and also remember the newsletter, misfits.news. Every Wednesday, a brand new edition of that comes out. And we have uh we have our pretty faces out there somewhere, too. Where's that at, Nor? Where where can they see our pretty faces? >> Our pretty face. Hey, you you can see it on Tik Tok just under marketing misfits and you can also go to YouTube uh for the long form. You can go to marketing misfits podcast if you want those 3minut and under clips. Uh that's on YouTube. It's marketing misfits clips. That's right. And other than that, I guess we'll see you again next Tuesday for a new edition of the marketing misfits. Until then, happy marketing. >> See you later.