#532 - Amazon Image and Listing Strategies

How split testing can make a huge impact to your bottom line. Where you should and shouldn't use AI for Amazon listing creatives. Your top 10 listing questions answered. This and more on today's episode. Hello everybody and welcome to the AM podcast. My name is Bradley Sutton and I'll be your host and this is the show where we discuss all things Amazon Tik Tok Shop and Walmart private label and how to generate recurring revenue streams 24 hours a day during the A.M. and the PM, hence the name of the show. Get it? AM PM podcast. And as a matter of fact, last week I went to Korea, spoke at a conference there, and piged out on my favorite Korean barbecue, pork belly. And even though I wasn't anywhere near a computer for a few days, I was still making money online. How cool is that? Pretty cool, I think. All right. Hello, everybody. Welcome to our workshop here. Let's go ahead and meet our guest today. if you guys can just introduce yourselves and you know what you guys do in the industry, some of your experiences. Start with Daniellea. >> Sure. >> I am Daniellea Bullsman, founder of mindfulgoods.co. Uh for the last eight years, we've been optimizing listings and storefronts for seven to nine figure sellers. Um we've worked with over 850 brands and today I'm going to be sharing exactly how we're creating creatives that convert for those brands. >> Awesome. Justin, next. >> Hi, I'm Justin. I am one of the co-founders of Pikfu. Uh if you don't know what Pikfu is, it's kind of like a digital focus group where we get real consumer human feedback on all your product decisions from product packaging to branding to design to main images, infographics, any of those things all within a matter of minutes. And so like Bradley said, we work with a lot of uh both newer sellers but also some of the largest brands who are testing, you know, hundreds of different products uh throughout their catalog. >> Okay. And Kamal last. >> Okay. Hey, so my name is Kamal. I'm the founder of AMZ OneStep. And we've been optimizing listings, doing productgraphy, videography for Amazon sellers, creating A+ content for, you know, over eight years now. And recently, I started a brand new AI startup. It's called Listing Optimization AI, which generates high converting Amazon images, you know, for you with AI in minutes. So that's that's my new gig for the last uh for past few months. >> Awesome. Awesome. Now, why they're all here today? Normally, we don't do these in person. And we do these these kind of webinars every month. So make sure you guys are on our list when when you guys want to get on these live workshops for a more in-depth workshop for our elite numbers where they have full presentations and we did some case studies w with with this product and and some other products where we can show you guys how to approach uh uh you know the task of making images uh split testing them uh doing things like a storefront and uh you know how do you how to uh put a listing together. I'm going to be talking about some launch uh aspects where a couple of launches I did in the past few weeks, I was able to get on page one for my main keywords in like 3 days. All right, so we're going to be talking about that workshop. So, those of you who are diamond or elite members, you guys will be able to get the replay of that. But for everybody else, we want to kind of give you guys kind of like a sneak peek of what we do in our elite program and give you guys a little taste of uh you know, the the the process and the kind of education that we give. So, let's actually start with a Kamal like I actually uh gave you guys this project of a real Amazon product that we're launching these uh children's workbooks um for a brand that's been on Amazon and Tik Tok shop for a while now. I sent you the physical copy, but did you actually take any images of it or did you do a photo shoot or or did you do all AI? What was your process and and how did you decide what that process would be? Okay. So when I first looked at the product, I instantly knew that okay this is going to be a blend of AI plus you know human work including photography because the reason why because that was a special like you know it was a cursive writing book and there's like a lot of tiny little details you know special fonts you know special animals characters so AI was going to was not going to get it perfectly right in the first code. I was absolutely sure about that and that's why I was like Bradley can you send me one product you know like sample so that at least we can generate mock-ups with the AI and then we can replace you know uh those pictures with the real real photos so that was the main reason we you know I ask you to ship that product to me uh but but the process looked you know like very hybrid you know I think the 80% of the work was done with AI maybe 85% of it like you know including the research coming up with the image ideas image concepts That's you know main image variations you know the whole ideation process was all AI and then you know the designer came in and photographer came in designer photoshopped the pictures you know from you know from the AI images put the real ones and also you know just just removed some of the AI you know slop and made it made it look perfect. So that was kind of uh the process uh in the background. >> Okay. Now, what's beautiful about the whole Amazon journey is there's no you almost can't say this in 2026, but no, there's more than one way to skin the cat. Like, you're not supposed to say those kind of things about animal cruelty and things like that. But the point is that there's not just one way to do things. And I love having contrasting opinions and workflows because we all can get to something that's that's sellable and something that's decent and approaching it in a different way. You know, there's some things where it's like, hey, there was just like one way that you probably, you know, you should be looking at search created performance reports and looking at what, you know, gave you good sales and like there's not much different ways to look at stuff like that. But other things when it comes to creatives, there's a many ways that you can approach that. So, so Daniela, when you approach something, uh, how how do you approach things like do you 100% always suggest photography as opposed to doing this hybrid uh, model? So, we love love love high quality photography that nothing will ever replace that. Um, but there's a lot of brands that are finding great success with AI and there's a lot of brands that are not finding success with AI. So, I think we have to balance both. One thing that you said earlier is you you might not even need Helium 10 to do your creatives. I emphatically disagree with that. Not to tout Helium 10, but I will just say data is the biggest factor in what we're doing. At the very beginning, we have to pull as much data as possible. And because Helium 10 has so much of it, we're able to grab so many different gaps, so many different opportunities, put all that together, and build a hypothesis. And that's really what tells you what creatives you need to be designing, right? It's it's if you go to Claude right now and you say, "Help me build an Amazon listing. Tell me which images I need to create. Here's my competitors." It'll do it like a half decent job at at analysis, but you're not giving it a north star in terms of what data it actually needs to be factoring in and considering. And that data is is the is the gold. That's the key. That's what's going to help you unlock the conversion levers. So for us, it's all about bringing in the data. Uh using AI and smart ways like rapid iteration is huge. Using AI u with tools like PFU is huge because we can analyze the data better than we we've ever been able to. Um and then using AI for scaling cataloges is huge. So, there's so many different different levers that we can pull now with AI, but it's about doing it in a smart way that makes sense for each brand. >> Yes. Now, let's let's actually show some of the pictures that we came up with. Hopefully, you guys can see this, those of you watching on YouTube. Now, the first thing we notice is these are all white background images. And so, you might be wondering, you can only have one white background image. Why did Kamal make four different versions? Uh Justin, why why do you think so? >> So that we could test it. Yeah. So that we could test it. Yeah. Against uh consumers. >> Talk a little bit about that. I think that's something I think all of us here agree on the split testing concept. But maybe this are not something that you know all of you out there uh have experience. And then for those who might not understand what that uh means, Justin, if you can explain that a little bit. >> Yeah. So you know, as as Kamal said, when you're working on a main image, you you typically have a lot of hypotheses of what's going to work. Maybe you're looking at what the competition's doing, but you don't know what's going to work until you put something out there. The danger of just putting something out there is that it's going to impact their sales. Even if you do a live AB experiment with man your experiments, you're still risking sales. And so, um, what all the advanced sellers are doing is pre-esting. And so, with a platform like ours, and and there's other ways that you can test it, is that you're testing off platform against your target audience. So, uh perhaps you're targeting parents with this one because you want your kids to learn how to do cursive. You could target parents and ask them which uh which of these variations um most appeals to them. Which one would they buy and >> parents who are also prime shoppers >> and parents who are yeah crime shoppers or maybe you're selling on Walmart and you want to do Walmart shopping. Um so you can do that targeting. Um the great thing is that it's not just the quantitative data. So they're not just going to select which one they like, but they're going to give written explanations why in their demographic information. So, you can read through that. You could analyze it and see, oh, they like, uh, you know, I was looking through some of the responses. Uh, I know that some of them like the call out that it was grooved. Um, and so that they could actually trace it. Um, people liked the the product in use, the the one that won actually showed the hand uh going through the cursive. And so there's all these small nuances that, you know, they all look great, but when you hit your target audience, like certain themes are going to resonate with them. And you know, a lot of times it's the layout. Sometimes, um, what we've found is that sometimes when the layout's like too elaborate, but you can't actually see what all the different things are. I think this was maybe one of your polls many years back where, um, it actually worked better to lay it out. So, you could see like the the contents of each different variation. So, there's all these small nuances that um, you're not going to understand if you just put it live and then maybe you see the performance dip and you don't know why. >> Yeah. So you actually want to hear from people like oh it's because you know you know the brown dog versus the black dog is less appealing and it's like oh people care about that people do. >> This is a very very specific type of product. So it's it's we can't just make like a hypothesis against a category although we do because we have like massive amounts of data but >> you have to start somewhere. >> You have to start somewhere. And for very niche types of products, this is the quickest, fastest way for you to get the data, change your designs, and impact your listing without it actually touching your listing yet. >> Yeah. >> For those of you looking at this, what I should have done this before I spoiled it because you I saw I screwed up, but which would you have done? Um, I personally thought that this first one >> might have been the pick because I I like ones that that have like a little call out. just makes it stick out on the page and and hey, it highlights the group pages. But which one one B, two or three that you guys can see, those of you watching, which would you have picked? Um and and especially those of you who are parents, you know, like there might be those of you single out there, no kids yet, maybe you have a differing opinion, but a lot of times when you run these polls, what you think as even yourself, as the brand owner, it's not the one that is is picked. And so what I did on this poll was I did do what they said. I chose parents uh as as a characteristic and I chose prime members. And this is uh the the information we got and it was this picture three >> what's actually won. It's interesting some of the stuff that is mentioned. All right. That I wasn't thinking about. It says like this person here says I like option C the best because I like that I can see the image in the upper leftand corner showing the examples of how the letter should be written. You see, like me, I know this product. >> Yeah. >> So, like, I don't need I I don't need that. Like, but then somebody who might be seeing this for the first time might not have even known until they saw that image that, hey, this is how it's used. I'm like, >> they want to see the paper inside, right? Cuz I think Yeah. The other ones have paper. And it's something about the hand drawings. We've done a ton of journal journal clients as well. And those ones always perform the best. The other one is um in the product image stack is a gifting image because journals are a highly giftable item. Why do the hands when you know they perform so much better? Because the process before you even you know capture the split testing is you kind of you know look at the sub simulation and see which image like does stand out. So what we like to do we don't we you know we we generate you know like maybe 10 15 main images before we you know short list the final ones which are going to go to Pikfu. So like 10 images and we look at which are standing out the ones with the hands or with the with the with the hang tag or with the product packaging they tend to break the pattern in this on the search results and you know it just it just stands out. So so that's why they they do really really well. >> Did you test that as well the against any competitors in the search? Not yet. But let's talk about that because it's not just about all right, I made four images. Which one should we go with now? Let's just go forward. What are some extra steps, Daniela, that we should take in this process of of of split testing? >> Our process is we're always trying to ideate as much as possible in the beginning. There's no wrong answers, right? You do this, too. It's like 50 main images if you want. You know, there's no wrong answers here. And then you take the ones that you think are the strongest. you would test them against each other and see which ones are trending in the right direction. Um, what you're looking for at that point is really like directional data so you can improve your images, iterate, make them even better, take the strongest ones, and really like narrow it down to like your top one or two. Then I would test those against some of uh, well, then I would test it against the baseline or whatever the client has. That's always what you want to do first. Make sure you're going to outperform whatever you have, period. Second, you want to test against your competitors. So that would be either the like SER type of testing in Pikfu which is its own thing or you can pull some of the just the main images from competitors and build your own. Um and there you want to see like you want a baseline to see how your original image was performing against your competitors and then you want to use the ones that already were proven through your your polling. So, you would take one of your top contenders that you think is going to perform well on Amazon before you upload it to Amazon, run that against your uh competitors, the same ones that you ran in that first poll, and make sure that you're you're getting more clicks, that you're getting a higher percentage of basically stealing market share from your competitors. That's that's the simulation that you're doing. >> Yeah. Now, we got save your questions for the I want to bring this out since we I might not have this uh screen up uh towards the end, but I know the answer to this question, but I'm going to let Justin uh reply here. How is a total greater than 100%. >> So, so the way we run the polls when it's a rank is actually um it's a preference test where you actually rank every individual. And so, um if you want to scroll down a little bit um if you have it up uh to the responses, you're actually going to see they they don't just choose I like option C. They're actually going to say I like C, then D, then A, then B. And so what this allows is that uh because if you have eight different options and everyone just chose their first choice, you might have 12 and a half percent across the board. It's not really useful. But there's actually nuance to that and a lot of elections use this and they they call it instant runoff polling where basically if you don't have enough winners uh if you don't have a majority winner in the first round, you actually discard uh uh you take the the votes from the lowest one and then you take their second place and then you apply it. And so what you're actually going to get is the most preferred option. And that's why they say that uh in areas that use this for uh political pol uh political elections, you actually don't have as much polarization because you actually have someone that can work with everyone because it's like, oh, this was actually everyone's most preferred. Maybe it was the B option, but it wasn't as polarizing as as the other two. And so the the reason why the percentage uh is that way um there's actually a little explanation on the poll results if you were to pull it up, but it's just the way that we're calculating um the the win percentage. >> Yeah. >> Yeah. Because it'd be different if it was just, hey, pick one. You're not going to get as much insight and you're not you get different insights. >> You know why this is great? Because let's say you you have a winner and for some reason Amazon rejects that image. Now you're like, okay, I'm going to go with my second, >> you know, if if this wasn't done, then you would have to re entire poll and like what the hell, right? So now if Amazon rejects the first one, you always have a, you know, fallback report. Yeah. Um, all right. Uh, come on. what is one of your top tips or strategies you're going to talk about today in the elite workshop and and just some you know maybe general summary in a couple minutes of some of the other topics you're talking about. So basically I'll be talking about you know how you can generate image stacks you know the question is like should you use use AI it's not a binary question yes or no it's like how much AI are you going to use right so I I'll be you know sharing your process I'm going to take an asen and we're going to uh you know hopefully generate the full image stack up to 85% or 90% whatever we can do right that's going to be live uh so the process is something it's going to look like this we're going to look at all the competitors we're going to scan their images we're going to scan their copy we're going you know look at the reviews your reviews your competitor's reviews you know we're going to analyze everything where we're going to see uh you know we're going to extract some insights using AI like what are who's the customer avatar who exactly is buying your product what are the to top pain points what are the decision-m factors what are the objections that you need to handle in your listing images uh you know what should be the price point and it's going to give us like the full image you know strategy like image im image stack point like Hey, these are the ideas that I'm thinking and then we're going to use, you know, that to build images, you know, using different image generation models. See which one, you know, does a better job. The ones we like, we'll upscale that to 4K. You know, uh we can potentially, you know, we can even using AI, we can push to seller central as well. Um and that's kind of the idea like you know pretty much building your whole image stack using AI. uh you know you can do like other stuff you know download search query performance you know give that to AI you know there's like a lot of things you know exactly how you generate images uh but the bottom line is that you know uh we will challenge oursel and build the full image stack using AI and then you know we then maybe Daniela comes in or Justin you know when we we'll do the live uh pickful poll as well we'll we'll see like why you know Daniela might share you know her insight It's like what needs to like what kind of human like you know feedback or expert opinion is necessary at the at the end of the image stack. Is it like ready to upload? Is it like you know not ready to upload? So that's kind of you know my talk. What's the difference of how you approach hey what goes into the main image stack as opposed to what's in the A+ content because it it's never been even forget about AI but just some people they would literally put the same images just in a different way from their image stack and that's what's in their A+ content but these are two completely different things. How do you decide the direction you're going for each? Okay, so I I think that's a great question. So most Amazon sellers they think of their Amazon listing as just one thing. What I like to break down your main image is a totally different process than your secondary images. And your secondary images is a totally different process than your A+ content. So for the main images, your you know my my my first does it break the pattern? Does it look interesting? Does it pass two second rule? You know, would it stand out in the search simulation? Like the whole focus of the main image is to get you the extra clickthrough rate, right? And it has nothing to do with the, you know, uh your second. the secondary images you do like completely differently. you do like you know image stat polls with your competitors you know that's where your pain points your you know different features benefits those things come out right that's where like the human psychology you know the CBR you know uh those kind of things come out the A+ content I think Daniela would probably be the best person to to talk about A+ content how I like to go about A+ content is uh you know to give some like brand elements to it first for example your first module should be the branded one uh you know you're talking about uh you can use the frequently asked questions you know uh to optimize for AI to overcome objections you can use uh you know um the cross-selling modules to to you know for upselling if you have other variations or similar products you know those are the areas where A+ content adds really you know like a unique value another thing you can use A+ content for if you have a unique selling you know like point and you can dedicate a full module on on that unique selling point thing I would I would only do the reputation in A+ content if something is super important. Otherwise, I would only add things which are missing from the image stack which could not make up at the top. But they are still important, but they're not as important as u you know uh the you know the image stack ones. Those ones would go over here with premium plus content you know you can decide which one goes to my fourth you know like when you in a carousel style images you know like that's where the priority number one you know important thing and then you scroll right then your second most important thing then you scroll right so that's how I would design a plus content it's a totally different process uh when it comes to a plus content or secondary which is an a you know >> and I think now everybody uh qualifies for premium plus content without even having to do those tricks. That was just something announced by Amazon the last uh couple of weeks. All right, so now you guys kind of know, you know, what uh Carmel's strategies are. Remember, the main part of the show is full Q&A with you guys. So, I hope you're writing down the questions. Don't put them in the the chat yet. We're going to have about 20 minutes. We're going to try and get to at least 20 of your your questions. Uh Daniela, what's a couple of your top strategies that you're talking about today that you know you can go ahead and share with the audience as watching? >> I don't want to give it all away. Don't give it all away. >> But I will I will say um one of the things that we are well we're pretty much known for two things. Um that makes us a little bit different in the industry. Um one we do split testing with every single project. We've been doing that for eight years with PFU. Um this is just awesome. Um and so what's interesting about that is we have so much data in our Pixu account and now with CLA we're able to analyze that across various different categories. So, what's exciting for me, and this is something I've been wanting for years, is that we can now analyze multiple polls at once, and we can analyze categories at once and understand what are the insights that are going to actually move the needle and that are moving the needle for our top clients today. And then we can serve that up and share um tactics, strategies. Um, and today we're going to share some of that um, in the form of the playbook for supplements that we made, but also some data points that I've never shared publicly before, which is going to be awesome. Um, and then the other thing we're known for is bringing in a direct to consumer style design into our process. So, um, it's it we serve a very particular type of client. We don't work with every brand. Um, we probably only work with about 10 to 15 brands a month. We're very selective on who we work with. Most of our brands are going to be eight or nine figure sellers. Um they have much larger cataloges typically. And so um with that comes a a different I mean the reason why they work with us because we have that very particular lens of how do we take the best practices that are working in the direct to consumer environment and bring those into a PDP on Amazon to bring their bring their aesthetic bring their brand to life on Amazon because I think that's that can be really challenging for brands even brands that have been legacy sellers selling for 10 20 years on Amazon they still struggle with this we see they come to us every single day and they're plateauing and their content needs to be refreshed. So, um, we're going to show you exactly how we've been doing that for brands and how we think through those best practices and apply them to PDP alongside the split testing. I think that formula for us is what is what really generates the results. Um, more than 80% of our clients are seeing anywhere from 20% increase in sales to 2500% increase in sales. And it's really the same formula every single time. Yes, the data points are different because we're gathering different data points from the split testing. Um, and yes, the design is different because every brand has their unique brand identity, but it's the same process every single time. And so, I'm going to share that process with everyone so they can build their own internal workflows around it. >> 30 second strategy, best strategy for brand storefront page. >> Oo, brand storefront page. Um I have two top tests that you should be running on your storefront. One is going to be testing your hierarchy. That is um basically um your navigation rather like how are people getting to your products and how are they getting to it quickly as quickly as possible. That's the number one thing they need to understand what your product is and how quickly they're getting it into the cart on the storefront. Um, yes, you have more attention than PDP, but it needs to be crystal clear what it is that they're getting. So, that there's multiple ways to do that. Um, so that's the number one test that we run. The second thing is, um, above the fold hero. Um, usually what you see is just Amazon products that like just look Amazon branded. And really, that's like one of the highest lever impact moves you make on your storefront is is creating a big huge hero section right at the top of your storefront and making sure that your best sellers are right there. Um, so you can create conversion right away. >> Awesome. Uh, Justin, your top split testing strategies you can share with the audience. Uh well, I'm going to um reiterate what Dan Daniela was saying where we uh we have a main image optimization playbook that we recommend and it does start by baselining your main image against let's say your top three competitors. So grab your top three competitors um from the from the SERs, take their images, test your image against their um their images against your target audience. Um not only are you going to see where you stand, but you're going to get all that feedback about why they chose those other options. So, it's like, "Oh, I like the box. I like the packaging. I like the hand." And all that feedback, you're going to take that, you're going to make the iterations with it that we talked about. We're going to test it just like the poll you ran. After you find that um that winner, you're going to revalidate it against the competitors. And an important nuance there is that you don't have to win this actually. >> Yes. >> So, you might have lost with like zero votes at the very beginning. If you even get five votes, 10 votes, you've made an improvement. You may not always win against your competition because maybe you're going against some monster brands or your product just isn't as good. That's fine. But you were just trying to increase uh what we can. We can we can improve the main image. We can steal some clicks. We can steal some market share. So we just like to show a relative improvement. Um that process is incredibly reliable if you do this like end to end. Um so what I'm going to be previewing um and what I'm going be talking about later is actually how to do this within Claude. Um, so you can use our P4 MTP with our connector to run all of these polls, run all these analysis, actually even do some image image generation of those variations. Um, and uh, the advanced stuff that I'm going to be showing is how to do this completely autonomously so that you don't have to touch anything. >> Um, and so that'll be for the the power sellers who have tons of SKs, they've got hundreds of SKs, they need to run this at scale, they don't have time to babysit it manually and definitely and not even within uh, Cloud itself. >> Awesome. Awesome. All right. So guys, we're going to talk uh in a workshop about PFU, but those of you who have Helium 10, you might not even realize you already have access to it. It's not called Pikfu and Helium 10. It's uh called Helium 10 audience, but it's run 100% by PFU. So those of 50% of you who are Helium 10 members, you could literally while we're on this call, go into your Helium 10 account and run uh a Helium 10 audience uh on on, you know, any of your your products. Now guys, as of now, this rest of this show is Q&A. There's already a few questions in there. We're going to get to it. I already got some questions here asking how to uh reach out to you guys. So, let's start with Kamal. How can people uh find you uh name your websites and and where's the best way to reach you? >> Yeah. So, you can contact me on LinkedIn. My name is Kaml Jits Singh. Find me on LinkedIn or you can just go to www.amz1step.com or www.listingoptimization.ai. That's where you can find me. You just click on the contact button. I'll see that message. Chassan. Uh same LinkedIn Justin Chen on there. Um obviously you can go to pfu.com pkfu.com or try it out in Helium 10. You don't need a new account. You already have your payment method in there so you can try it out. Um if you do come to pfu.com we do have a free onboarding. Uh so you can try it out. Uh there's like a mini poll that you can run for free. Um otherwise feel free to email me.com. I'd love to hear how you're trying to test. Um if you're trying to do things autonomously, I'd love to hear the different stories. Danielle. >> Uh, you can find me on LinkedIn, Daniela Bolson, and my website is mindfulgoods.co. >> Okay. All right. Let's go ahead and get into the, uh, questions here. Let's go rapid fire on some of these. This actually is for Justin, but to make it more real, I'm going to have somebody who doesn't have skin in the game answer this. Uh, Kamal, why would you use Pikfu instead of just running Amazon's manage your experiments a uh, AB testing? >> Great question. Okay, I think number one, I think it's it's it's fast. You get the instant results. You don't have to wait for weeks. That's number one reason. The second reason is, you know, like Justin mentioned already, you don't want to take a risk with image which is not tested already. I like to call it, you know, a practice game. So, you want to win the practice game before you actually play the live, you know, like the real. So, so don't skip that. Uh, so that's the second reason. Third one, you're going to get feedback like why people chose that image and or why what was the reason. So you're going to extract so many insights that probably you were not even thinking about and you would be surprised to see like what the hell like why did 50% of the people mention about this thing that I did not even notice. So it will give you so many insights and now you can take that you know feedback and now that's something we call it iteration process or or or or the revisions process. If you don't do that, you know, that means you're you're just taking risk live and then you know just then then you don't even get the feedback and you're why it's not working and then you have nothing to make changes based on you know you're just playing blind which is very risky for a business and there's like many other reasons uh why you should be used to using fit proofs before you actually test live. So but these are the my top three reasons. >> I'll add to that really quick >> real quick. Yeah, I get to all this. >> Here's the thing. Because both of us have been running polls for eight years. We have massive amounts of data, right? For us, it's a non-negotiable. Like we would never do a project without running split testing. It's just like it's just something we would never do. We did an analysis on a bunch of our poo poos. And we correlated the data, which I'm going to share today, between brands that ran their split testing and then did their Amazon experiments afterwards. 70 75% of the brands that did um P that won in their Pikfu testing and put those those versions live to Amazon experiments, 75% of them showed an increase in sales. So there's an argument that people make like why would I use a third party testing tool? It's not real Amazon shoppers. It is. They're Prime members. You can select Prime members and 75% of your results will carry over into increase sales. So >> yeah, my data shows it's 80%. So uh I have you know we mentioned an NCB in a few few times. So so two insights that I got if you split test your main images there's 80% chances that the new image is going to win. That was one insight that I got got from NCP. The second one the difference between you know average main image and a great main image is about two to five times the CTR. These were the insights that I got and this is a mind blown out. >> Let's stick let's stick with Justin here. We're testing competitor images. I think Danella was talking about that. Yeah. Are we talking about product placement, positioning the image being the same as a competitor? So, it's also about when you're split testing against the competitor, it's like you're not just split testing the image, but actually kind of like how it looks in the search results, right? You unpicked food. Can you talk about that real quick >> about product placement? >> Yeah. Oh, it's not just the image. It shows the number of reviews and and stuff like that. >> Uh, well, I'll take it both ways because I'm not sure how they're talking about it. First of all, like within a main image, um you'll be surprised at how much like position left versus position right like makes like sometimes people just like to see the product position this way or this way or you know certain angles. So, you know, aside from doing all the little tips and tricks like even just that kind of product placement can actually make a really big difference. Um but then yeah, you do want to see how it looks relative to um to the uh actual uh SER. So we do have a uh we do have a SER test where you can um mock it up into uh you know a mock environment or you can actually just take a screenshot um and do like a click test where you will generate a heat map. So there's a lot of different ways that you could test out how it actually looks um within the context of of the other search competitors. >> All right, going to Kamal now. Uh our friend from Greece here says, "I've got a question about Amazon main images. I've noticed a increasing number of listings are using labels or graphic elements in smart ways to make products stand out even though main image rules are traditionally quite strict. Has Amazon become more flexible with this or is it still considered risky from a compliance point of view? And we could probably talk 10 minutes but 60 seconds or less. >> Okay, so we I get that question a lot. So Amazon has you know there are some categories which where Amazon is like super strict. it won't even allow anything you know outside you can't really do it it has to be pure white background product only and there are some categories where I've seen like lifestyle images being used as a main image so it first I would check like you know if your competitors are using that I think you should use that too but if everybody's using those strategies and the whole purpose of the main image is to break the pattern and not do not do you have to do the opposite of what what you know what everybody else is doing because we were talking about search simulation that's where I would I would check. Uh so that's what Yeah, I only have 60 seconds, but but but yes, there's different categories and uh I think that's one thing which is tough. >> Have fun, play with it. The Amazon will slap you on the wrist. >> If they're doing it, you can do it. >> Yeah. Yeah. There are some things guys on Amazon you don't mess with like like fake reviews or we're doing crazy things like like that. Just like don't even dabble in something that you can consider a great area. There are other things where the worst that can happen is like not bad at all. Like where Amazon might say, "Hey, this image is not compliant. Actually, it doesn't even suspend it. It might just uh it's why you should have like a white background image as like maybe your seventh or eighth image because and sometimes if they don't like your main image, they'll just switch it automatically without even letting you know and like you won't even get a slap on the wrist." So, so just you know know which things that you can kind of play with and other things where are non-negotiable such as reviews. Uh Daniela Valentine says, "How much does having a good storefront influence a product purchase? Does it noticeably increase the CVR or it's not a determining factor?" >> Uh it absolutely noticeably increases the CVR. And we've tested this on brands that actually are not driving any traffic whatsoever. They're just their storefronts were just sitting there and almost every single time their sales went up. So that's that's interesting to me because um it's it's just with little bits of effort you can make meaningful improvements whereas like on PDP it's a very competitive space you know to be changing your listing it's it's it's it's so many different things you have to factor in whereas on your storefront you have way more control and there's already organic traffic so go check your reports and see how much traffic is landing there and um play around with it you know >> uh Hosia says hey uh Amazon Listings typically have less text, have clean aesthetics. In the age of AI, with Rufus driving 20% of guys, don't drink that Kool-Aid. Rufus is not driving 20% of the sales on on Amazon. Rufus is great for a followup. Basically, the kind of things that you would do anyways. Like, if you were as a buyer on an Amazon listing, what would you do? You would look at all the images. You would read the details to see what it's compatible for. You would summarize the reviews and look read the reviews. Rufus is helping with that. But if you if you if you're trying to say that, hey, roofers drive 20% of all sales through search, guys, that's not what people are saying um out there and and that's not what Amazon is is trying to say either, is it? But that being said, yes, having more text in your listening for Alexa, formerly known as Rufus, it can read that. I've tested it before where there's a question that maybe Alexa asks or you ask Alexa a question and your listing doesn't answer it and you put it in an image or you put it like in a bullet point or something within 10 minutes Alexa will be able again what I the artist formerly known as Rufus will be able to change their answer and so it's really amazing how if you notice a gap with Alexa or Rufus that making these changes to your listing can instantly change it. Whereas in the old days, you were just talking about SEO, you make a change to your list, it might take like a week for for you to see results. But Alexa is like within minutes. So that's something good. But in this age of AI, are you guys doing anything differently like putting more text on the images either in your A+ content or having more infographics because of AI or or are you kind of doing the same thing? >> I I think a little bit different. It's it's we we say that it's text but AI does not really need to know the text to understand the context of the image. To give you a couple of examples, one we used, you know, like we didn't put anything about perfect gift or anything in the text, but we had a gift box in the background and then we, you know, if you use listing optimization.ai, we have a rules and cosmos section now known as Alexa. Like it analyzed as a giftable. Yeah, this is, you know, it said yes, this is this question is answered about the giftability. It's a giftable product. Second one, you know, we had that small icons, you know, it was a it was a product which was made for school and then parties and you know and then the question was like is it for adults or kids only like just from the school and the parties it it understood that yes it's for adults and kids. So the how you know AI understand it's not only because of the text it's like it's looking at the image semantically. So you want to you want to make sure your images are like you know semantically optimized as opposed to just putting information in terms of text only. So that's the beautiful thing about AI. >> Okay. Quick one for KL here. How do you photograph your base image before inserting AI tools? So that that uh grooved handwriting thing I sent you you then did AI but obviously there you know didn't just magically appear in AI. So what do you do for that? >> Um ju just just take you know um if if you know that you're going to take the real pictures it does not really matter. take a picture from your cell phone like the way AI will automatically rearrange things for you. It's not going to be perfect but uh but that's where you can start and that's where you can start once you're like you know happy with the concept then you can take a picture from the exact angle and then a then ask AI to replace that picture you know from the concept. So don't just get into, you know, uh that mindset where you have to take a perfect product picture before you begin with. Use AI. Let it mess up your product. That's fine. And then you glaze a product. >> Quick one I'll do here. Can Rufus Can Rufus or Alexa read the text production uh text description that is hidden once we add A+ content? In my testings, both SEO and uh Alexa can still read that. So like if you have A+ content, which everybody should have, you still should have your description uh in the background. But don't just keyword stuff that because if something happens to your A+ content, the listing reverts to that description that you have there. So if you're just like keyword stuffing the description because you're like, "Oh, nobody's ever going to read this." You might have a bad experience if something ever happens to your your A+ content. But yeah, last I checked, uh, Alexa and SEO, both indexes, parts of it, not 100%. All right, let's see here. Uh, there there are some questions asking about will there be a replay? Yes, we're going to have this on the AM PM podcast YouTube. And then also, if you want the replay of our workshop, normally that's only for elite members. Like elite members can attend. And um but if we're going to open it up just for this one, the the 7-hour workshop that we're doing later today, anybody who's a diamond member on this call uh will be able to get access to their deep dive uh workshop that we're doing later today. Um I'm going to throw up a quick coupon code if anybody just wants to join for like one month just to be able to get this. It's 20% off. You can use the code SSP 20. SP20. All right. So use that for uh to get 20% off the diamond plan. Obviously, Kamal has listing optimization.ai. Um, Daniela, when you're you don't use AI too much in the in the production, but do do for your mockups, do do you use like chat QBT or something um uh in order to like mock up your kind of like storyboard before you do the the production? And if so, which AIS are you using? >> So, what's interesting is um the storyboard we're primarily just using Figma and and you know, it's mostly text based at this point, not visual based. once we get drafts back from our designers um that's where we're really like rapid iterating with AI um and for that I'm typically using chatbt although I feel like every week it's changing and when a new when a new model comes out I'm like oh I want to see how it performs against this so I think it's we're all in this state of constant learning you know I don't think there's any one way to do this right now everyone's constantly growing and learning and that's what's so exciting about it um yeah >> after optimizing your photos and listing. How do you know it's ready? I find myself tweaking my listing and photos a lot. This is my first listing. I want to do well. Uh Dea is is suffering from analysis paralysis which we all have gone through. So who who would like to talk about this? >> Maybe I can take this one. So uh yeah, if you follow the process right you know go through you know uh like let's say pick food polling first and then go through manager you know manager experiments and then also keep tracking your you know CTR and CVR. So any change that you make, you know, it should reflect in something. If you're making a change in in your main image, your CTR should either go up or go down or stay flat. So, so I would say just, you know, keep testing and keep tracking. There's no right or wrong answer here. You just have to, it's not a one-time task. It's a process and you should never stop this. Um, so yeah, just, you know, for example, if you're if you're worried about this change will make a difference or not, yeah, just just test it out for a week and you you'll find out. So you can always go back. That's a good thing about it. >> Last question of the day is about video. First of all, in Pigfu, can I split test video at all? >> You can use it for that. >> Yeah. So you can upload video. You you don't even have to split test. You can get open-ended feedback on a video. So maybe it's your existing video or you have a a version that you want to get feedback on. You could ask audience to watch it. Um and you don't want to just say like, do you like this video? A lot of times you have a video for a purpose. So you know, what is the takeaway from this video? um you know is it because maybe it's conveying um security or um sustainability or whatever it is you probably have a purpose to it. Uh the other way that people test uh as they're creating videos they test the different aspects of the video. So they might be testing storyboards or voiceover actors uh background music all these components. Um if you wait till the very end to test something then you're you're getting feedback on all those micro decisions that you've already made. But if you can get feedback along the way, then you're, you know, you're making the best decision. And so like that final product is going to be a lot more put together. >> And last uh question along that line, um about AI videos, um video ads, are you are you doing anything AI as far as the video advertising or that's not literally what you work on or or creating any kind of like unboxing video or something with AI? >> So we've always done UGC. Um, we've gotten a bit into AI ads. Um, I think all of that is changing right now, especially with the law that passed in New York where your content, if you're running advertising, has to be labeled that it has synthetic AI humans. And so, the way that we're going about this is we're treating uh we we are choosing to treat PDP as an advertising because I don't know what the gray area is legally. Um, and so we are going to be labeling all of our content um whether it has AI or not, we're going to be labeling it so that there's transparency in the process. I think that's important. I think it's a it's a good step in the right direction. I think that um there's going to be more laws that follow this and this is just the beginning. We do need boundaries around AI. So, I think that that's a positive thing. Um uh Amazon has its uh creative studio. I think it's just called Amazon Creative Studio and in there you can drop your product. Um you actually don't even need to drop anything in there. You could just tell it um go find my PDP and uh make me a video about my product. and it makes incredible video content. Um, when we ask it to make specific snippets, um, it'll make a few different versions. And, you know, sometimes there's a piece of it that's not great. It's a little sloppy, but there's little bits of like one, two, or three second clips from almost every single production that it puts out that's just amazing. And so, what we'll do is we'll take little snippets um, and and kind of stitch them together in in a post editing software to make an ad from that. And so there you're going to have more control to add text and all that kind of stuff, which Amazon Creator Studio does have that as an option. It's just for me it's a little slow and so I like to move it into other software. >> All right, thank you guys very much for tuning in. Thank you guys for coming from Panza, Culver City, and Peru all the way to present this and we'll see you guys next time. Bye-bye now. Bye.