028: Vanessa Hung Unveils the Art of AI-Driven Product Listing & Advertising on Amazon

Speaker 1:
0:00

Welcome everyone to the brand fortress HQ podcast. I'm your host, John Stojan, and I have one of our other co-founders, Matt Atkins, with here here with us and I'm really excited about our guest today. Vanessa Hong is the CEO of online seller solutions. Vanessa is a dynamic entrepreneur reshaping the landscape of e-commerce with her passion for Amazon and her focus on operational efficiency. Her company specializes in optimizing catalog inventory and account health, along with recognizing that critical role of efficient management in being a successful Amazon business. Vanessa, welcome to the podcast.

Speaker 2:
0:34

Thank you so much. I'm super excited about our topic today.

Speaker 1:
0:38

Yeah, absolutely. Well, before we dive into that, can you give folks a little bit of background on? You know kind of how you got started in Amazon e-commerce and you know where you're at now? And then you know I know we talked before we hit record here about Amazon and AI. After that we'll dive in.

Speaker 2:
0:56

Yeah, perfect, and I guess you asked a question and I always get asked that question. It's just like to see the credentials right, like, okay, I started in the platform in 2018. Sorry, yeah, 2018. Working as an e-commerce manager there, I had the privilege of managing three large accounts with massive catalogs and I'm a nerd, so I basically dive into the back end of Amazon Not really not the sexy side. So I I specialize in what I call the unsexy side of things, which is the back end of the catalog and the account, health and the algorithm and the inventory and things like that.

Speaker 2:
1:37

And when I started going out into the community and reaching out to people and meeting a lot of like providers and events and stuff like that, I realized that there was a massive, massive gap between what it was talk about on the launching side of products and the PPC and how you grow brands, and nothing was talked about on the managing part of the business, like how do you fix an issue, how do you communicate with Amazon? How do you, you know, manage the inventory properly? So that was like a big opportunity that I saw. I started, you know, working with clients, freelancing, consulting, then founded the agency and I've been doing the agency for about three years and a half now and it's amazing because the agency allows me to work with every type of seller, every category, every size, and I'm super privileged to see like the forefront of every single issue. So with that data I'm just kicking out on new ways to solve stuff and prevent.

Speaker 2:
2:40

I'm also very heavy, like heavily invested, on the prevention side. I think if you build a foundation, a strong foundation, in your business, you won't need to fix a lot of issues on Amazon. So yeah, basically that. And I'm also a community ambassador for Carbon Six, which is an ecosystem of tools, and yeah, I'm just an Amazon nerd.

Speaker 1:
3:04

I launched several brands and you know somewhere good, somewhere bad, but yeah, Okay, well, I think that that segues really well into our topic today, which is talking about AI and Amazon. We know that AI is a super hot topic and obviously you know Amazon is working on incorporating more and more into AI, into how they do listings and a lot of other things. So tell us about what you're seeing in the ecosystem right now.

Speaker 2:
3:33

Yeah, so completely disruptive, I will like with the news that we have seen so several things happen in the platform, for example, the summary and the reviews. So on the front end of the detail page you now see a summary, like AI summary, on the reviews. That's one thing that implemented in the front end. Then they released a news I don't know if you saw that. I think it was probably a month ago, oh yeah, but the beginning of February.

Speaker 2:
4:00

Rufus, which is their personal shopper, AI, basically now when you log in into your desktop or your app and not all all customers have it but you will eventually get access to this kind of bot where, instead of looking for a keyword, a specific keyword, you can look for products based on prompts, and your prompts are, for example, let's say, hosting a birthday party, what should I get? So the AI will prioritize all of the things that he thinks, or the AI thinks that you need to get for a birthday party, like cups and napkins and plates and balloons and things like that. So when I saw that, that to me was just so disruptive to the way we are building our listings, we're building our catalog, and I got super excited. I think for a lot of people. This is scary and I understand, but I always try to look at. You know, Amazon is an ecosystem that will always change and innovate. So if you are afraid of innovation, you better, like, go into construction or something like that. You know like.

Speaker 1:
5:22

Yeah, amazon moves fast. So if you're not prepared to move, I mean, and the nice thing and I hundred percent agree, and I just, you know, want to double click on what you said there which is, in the time, over the years, there's been nothing but change on Amazon and there's gonna continue to be changed. There's two ways you can look at that. One is oh my gosh, why do I have to deal with all this change and this is such a pain in the butt in order to keep up? Or you can look at it and say okay, probably half of my competitors are not going to implement anything based on this change. So just by you know, adapting to the big shifts on Amazon, you're already ahead of half your competitors and it creates opportunities that you wouldn't have had before. So I think that's such a great point.

Speaker 3:
6:06

First, first movers advantage is real. I mean when you start, when they launch a new ad type and you're the first one of your competitors to use that new ad type. I mean you have prime real estate and it's the same in this case, like soon there's gonna be ways that you can bid on ways that people are now searching for on Amazon. So the first person to figure that out and to be able to target in that way and set your listing up in that way, you're gonna be light years ahead of your competitors.

Speaker 2:
6:31

Yes, so that was to me. When I saw that, I'm like sign me up, let's crack this code, let's do it okay. So then I I Because it's so funny, and I say this, and sometimes people like laugh and say like that's not. That can be possibly true, but I do read Amazon policies before going to bed Like that's the thing, that's it, that's a ritual. I'm like let me see what's up.

Speaker 1:
7:00

One white Sleep or is because you're interested?

Speaker 2:
7:05

No, I mean, you know the reading before bed is just good for you, whatever. But you know, yeah, sometimes they put me to sleep, not gonna lie. But there is one white paper that I found recently. They just publish it I think was at the beginning at the end of January, beginning of February. It's called, so I'm gonna read it Cosmo, a large-scale e-commerce, common-send, knowledge Generation and serving systems on Amazon. Super technical, you don't like? Don't worry about the words, I'm just gonna, you know, break it down. So Rufus is is the name of the chatbot that customers will start seeing in their apps, are in the desktops and everywhere in the ecosystem. That's one piece which is the front-end customer facing.

Speaker 2:
7:53

Cosmo is what we could equal to the a9 algorithm and, basically, what cost matters. I think it's an addition to a9. I don't think it's a substitute of day nine, because they nine still exists and it will keep existing, but it basically what it does is technology is that it goes deeper into the Understanding on the intentions and the customer profile, and what I mean with that is based on their Super fancy mathematical research and all of the things that they could do in the back end. They could know the target audience, they can understand the, the profile on the customer, but also at the intention behind a specific purchase. So, as an example, and how the a9 is built, because we need to understand what's the difference, day nine, we are we, it was built and we were playing in this game for such a long time. 2024 is a year where, if you keep playing that game, you go, you will lose, but you, we basically sell keywords. Right, we don't sell pros, we sell keywords. So let's say that you have a coffee mug, you're selling coffee, like that, that's your keyword. Or best gift for mom, or best addition to kitchen, whatever, whatever keyword that you think of when you could purchase a coffee mug, that is what you're gonna sell. And the only way for you to sell it right now is if, if you index and rank or advertise under that specific keyword. Totally makes sense. That's a game that we've been playing in the back end. They nine is built to recognize Async or a listing based on a function called is a. So is a coffee mug, is a gift, is a tumbler, is a Accessory, it's a. So that is basically matching keyword to product. So it's a. And for books, is author of. So they match the author to the name of the book, that's those are the two functions.

Speaker 2:
10:03

Based on what I read on the custom, on the Cosmo White paper, what Cosmo is doing which that's why I think is an addition today, nine, not a complete rebuild or Subsided is that now they're targeting more things. They're trying to understand what's the reasoning and what's the Kind of intentions behind a purchase. So they now their relations are 15. For example, use for function, so that the mug, the coffee mug, is used to drink coffee in a you know cozy night or Drink hot chocolate on the slopes. You know it's directly related to a function, not a keyword anymore, and so let me keep reading.

Speaker 2:
10:56

Use for event, for example, and they do that in the example in the paper. They recognize that the person is got invited to a wedding and now that the the intention is to buy a Close for the wedding, so they put the intention behind that and it's used for an event. So now the algorithm knows that that dress is used for weddings, so that's a particular use. It's not necessarily a keyword that is attached to the dress, right then use for audience, and the audience part is the most Interesting thing, guys, and this is where I believe I'm kind of predicting that the turn that we're gonna take is in that way. Is that, as I mentioned before, we sell keywords? With Cosmo and with the AI and the personal shocker, I think sellers will start focusing more on audiences rather than keywords. So if my audience is a 40 year old mom with three kids in the suburbs of Chicago, then that the algorithm or the system will give me the details and which is something that they have started giving us in brand registry and brand analytics and all those stuff, like it's a demographic Right, not in so many details, but we now start seeing that. Why? Because we're turning into selling intentions and selling usability. So now our copy is getting analyzed by a system on a semantic way. It's not match to match, keyword to product anymore.

Speaker 2:
12:35

Other examples are capable of use as use on use in location, use in body, so huge things.

Speaker 2:
12:44

So when you read the white paper, they have a bunch of formulas and they're so fancy, but what basically does is each time and they start collecting information because what they're trying to build is a database of common knowledge they are trying to assume certain things so they can show, in the same way that the A9 now shows, the best, the best road for that specific keyword. They want to start showing the best road for that specific intention. And another fact that another part of what the common sense that they're building is. They so, for example, and they put the example in the A9 right now, if you buy an Apple Watch, the system recognized that you wanted a watch. But sometimes you don't buy an Apple Watch. It's because you want a watch, it's because you want an intelligent thing, or you want to play music while you're running, or you want to do things like that, or you're just an Apple fan.

Speaker 2:
13:45

They are trying to get that information out of the with the Cosmo analysis and the common knowledge and understanding the customer profile, which also kind of scary that Amazon is really gathering a lot of data on you, especially in other areas that are not necessarily Amazon marketplace things like Amazon Prime. If they see that you only watch sport content, then they already know that this guy is a sport fan, right, and they're building the intentions to get now a new algorithm. So when somebody comes in and say, like I have a birthday party, they already know okay, this is the mom in the suburbs with three kids. We better give her balloons and we got to give her like party supplies that are for kids or show her those type of things, and the results, in my opinion, in the future I cannot even imagine. But in the future, will we really tailor to audiences rather than broad keywords searches?

Speaker 1:
14:53

So let's pause there, because there's a lot to digest.

Speaker 3:
14:56

So much, so much to unpack.

Speaker 1:
14:58

I think, to understand as an ad, more of an ad nerd myself.

Speaker 1:
15:03

So what I heard you say and let me know your thoughts on this is that where we're at right now is almost like if you had an exact match ad, whereas if I'm advertising for coffee cups, so my ad is going to show up on every time somebody types in exactly coffee cup, but if they put in coffee tumbler for travel, I'm not going to show up there.

Speaker 1:
15:26

So right now, the A9 algorithm, like you said, is very focused on keywords based on what you're seeing from this white paper on Cosmo. Where you see it going or Cosmos, I think, is the name of it right when you see it going is more of that intentional based search, not only based on what they're typing in, but then also based on all the other data that Amazon has on that customer. So it might be coffee cup for ski trip. Well, now, if there's a coffee cup that maybe has a clip on it and designed for being outdoors and really intended for that purpose, with this Cosmos in theory, who weight that more than another product? That may just be a standard coffee cup, for example.

Speaker 2:
16:20

Yeah, and this is a conversation that I've well, because this is pretty new and this is my first time speaking publicly about it, and since you know a lot of advertising, the first question that came to mind for me was like, if this is true and they're going to do all of these things automatically, for the customer and the seller will need to build based on intentions and audiences how do you advertise? Because they're going to kill their biggest business, which is Amazon advertising, the biggest revenue driver in the marketplace, because Amazon advertising is not built for this. You don't have. It's not like Facebook. You know, when you do ads on Facebook, you go into the demographic. Right, I want this specific person to see my ad. We still don't have that yet on Amazon.

Speaker 2:
17:16

So that was one thing of like. Oh, I don't know. I don't know how they're going to merge those things, and I think that's the reason why they haven't released the personal shopper Rufus to every single customer in the US, because the moment they do that, basically nobody will start seeing the the your sponsor product ads Like. That's it. That's another game.

Speaker 2:
17:38

So I don't know no-transcript.

Speaker 3:
17:40

So there is a way that you can target based on demographic and it's with DSP, and what this probably becomes is they're going to start integrating DSP a whole lot more into or make it a lot more accessible. So, like Spontar Display, they're already releasing a lot of really cool things. With Spontar Display in terms of how you can target that, it's almost like a little mini DSP. So if Amazon knows now and is looking at demographic information or the intent, as we talked about, they already have that data, like they have the first party data of what we're purchasing with our credit cards. They're already allowing you to target people based on that. So now, now when they start releasing Rufus and they're going to start learning how people start searching for things because our search is going to how we search for things will change naturally when we have a new tool that starts to understand things a whole lot better.

Speaker 3:
18:31

So, like I mean, you see in the news all these different things that AI is touching, and for people like us that like to nerd out on this stuff, like I'm, I'm excited about figuring it out, decoding it and then figuring out how to show up in front of the right person. But when Amazon gets this right and you know that they'll end up getting you right. Imagine getting in front of the person that's getting ready to go on that ski trip and we know that because they bought a coffee mug and they included ski trip. Well, now I have all my other things that I know that they're going to need for this trip. I'm going to get in front of them at the exact right time when they're looking for that exact right thing like that. I mean, it's so exciting that the possibilities are endless of how you can target people based on these things.

Speaker 2:
19:10

Exactly, and on that side, that's amazing. I mean, I'm just waiting for that to be the reality. And then on the other side, which is my expertise, I don't know anything about us, but on the catalog level, okay, so let's unpack that, because that's probably the biggest question of like holy you know how do I adjust my whipping.

Speaker 1:
19:30

Now for new, essentially completely new way of product discoverability.

Speaker 2:
19:37

Yeah. So a few other things in the AI the way the algorithm sees your listings Now, with their knowledge and analysis that they're able to do, they don't only going to match the keywords or at the moment and this is happening right now this is not future cost model. When they get implemented, I don't know like well, we're gonna talk about data, but right now the system understands your product based on the keywords and the semantics of your copy. So in the AI I will not only see this is a coffee mug, but because of the semantics and because when you put in the copy and also what you have on your images Now, they can make assumptions of what your product is about and what is the intention and what's the lifestyle and what's the use cases and what are those things. So, first thing that you need to do well, a few. But the first thing is understanding or forgetting completely about keyword stuffing. So if you repeat keywords all over your listing, that's kind of over like you can start transitioning that opening space on your title or in your bullet points, because that will be redundant. Okay, the other thing that it's very critical and I know this because I'm on the other side of trying to solve the problem, when I already happened, is the system now analyzes images Okay, so, based on them and also cut.

Speaker 2:
21:15

Well, no, it's another white paper, but there's another white paper on the AI that analyzes the images on Amazon and how they analyze it. They don't only see what are the components on the images. So if you take a picture of me right now, it's not only like women darker glasses, microphones, no, no, no, it's like this is a person talking and it will make more assumptions of what's happening in the environment. It's not only a descriptive thing. So, since that's going to already getting implemented in the system, one thing that you need to do is you need to go and reverse engineer your images. So what I mean by that is you're going to take your images, all of the ones that are on your listing, and you're going to upload it to whatever AI you like.

Speaker 2:
22:02

I like chat GPT. You will update it to chat GPT and you will say, like, explain this image to me and the AI will tell you exactly, based, obviously, on their learning language model, which might not be the same that Amazon has, but it's going to be pretty close. It will tell you exactly what it is. So, when I presented about selling to the AI, kevin King took the content and he tested on his products. So he basically did that reverse engineer thing where he upload his image to the AI and he was selling like portable glasses, like you could put in a case on your phone, and it was super cool. The AI was saying that the thing that was in the case was a keychain no glasses no glasses.

Speaker 2:
22:53

So an example of why I tell you this is breaking things apart in the catalog. We work with a client that sells greeting cards and in the greeting card she has a red car because it's for a birthday. It's for like a boy's birthday, three year birthday, whatever, and it's a red car. So the system Amazon categorized that product on their toys and games because, based on the image, the system was reading that that was a car, not a greeting card, and that was just like what is going on? What is going on? That was my first thing of like.

Speaker 1:
23:32

Something is changing drastically in the back end, so chat epit was reading it as a card, or you're saying that, amazon, so it was putting in the category of toys and games, because it's like hey, this is a card that goes along with a card game, not like a greeting card. That's so interesting because I, you know, I've definitely had clients where they're super frustrated because all of a sudden Amazon puts them at a different category, and I think that this is a good example of how that can happen.

Speaker 2:
24:05

Yep Images Huge, huge, huge source Another client that we work on what work with. In his listing he had party supplies actually, and it was a combo. It was like the sign on the cups, on the plates, on the napkins, on the all the tentacles. They're all of it. So he was under a category that was party supplies, but Amazon categorized him under home and kitchen or something like that. I don't remember if we well, no, sorry, under toys and games, and under toys and games he was under party sets, which which definitely is a party set, that's a party set.

Speaker 2:
24:47

So we appeal and appeal, and appeal and appeal until we got to a point where the representative again this is another queue that I got where I'm like, oh, something is really changing. The Amazon rep said this is getting analyzed or, you know, understood by the system as a party set. Because all of the things that you have, if you want to change this to the proper category, you basically need to change your product Like your image. Is that so on one side, guys, I think it's gonna be painful for the sellers that are not analyzing and reversing, engineer this and making sure their images are clear, not for the customer. Not for the customer they're selling to the AI. So you need to make sure that you are properly getting understood by the AI. But also, this will prevent a lot of people getting listed in categories that they shouldn't supposed to be. So for the bad players, I think this will be a horrible thing that's gonna happen to them. And for the sellers that are doing good and they're actually you know they're in the right category, they need to double check the images. So that's a big thing and if they and if chat to PT and you can also also test, I don't know. I know Kevin mentioned the AI that he used to reverse engineer the images. I don't remember the name, but I could get it back to you if you're gonna put it in the notes, and I guess you could work with any the thing here, and that's another part.

Speaker 2:
26:20

The third step that you should take is the different AI, so Gemini and GPT and part and the other the other one. Those are built. It's the same concept, but they're built in with different rules. So chat to PT rules might not be the same rules that Amazon have. So, for the third step that you should use is start create it.

Speaker 2:
26:46

When you go and create a new async from scratch, right now you have that little box where the system can generate a listing with AI, right, so you could either put a prompt or you could upload an image. And what I like about this is when I upload an image on the on Amazon, amazon is giving me their analysis on that image and the title that that probably should have. So I tested this with a standing desk the one that I'm using right now, actually and I'm like, oh, let me see, and a bunch of the things that they were saying they were true, but other stuff. It was like oh, this is interesting, why are they taking this specific, like? How are they taking this piece of the content from that picture?

Speaker 2:
27:35

So this is the things that you should reverse, engineer and analyze, making sure that your async are actually understood by the system in the way you want them to be understood, because I don't know if you, as you said, johnny, if you're selling a coffee mug and your target audience is people in the outdoors, right, like campers and like skiers and snowboarders and whatever hikers, that's probably a different way you will build images and a different way for you to create the copy. Again, we're going back to the semantic. This is not keyword stuffing. It's not coffee mug for outdoors. I mean, yeah, part of it, but you need to explain what it's about and also in the lifestyle images. So those sort of things.

Speaker 1:
28:20

A quick question on the, because I think that's such a great tip on the background of so, with Amazon using their AI to create the listing based on your image, so you upload that image and then Amazon gives you title bullets, those types of things, so you can really see how is Amazon interpreting that image. So could you basically just burn a well, you know, in a sense a UPC to create a draft listing and then use that as your updating your live listings and launching new products, to just put images in there to see what it's giving you back?

Speaker 2:
28:57

Yeah, I don't even use a UPC. I didn't burn a UPC, I just put generic as a brand and no, no product identifier. Gotcha, most accounts most accounts you can create generic.

Speaker 2:
29:09

So, yeah, no problem, you don't need to do that. But on the other part and that's the other nerd stuff that I was playing with was I went into the AI and I started playing around with creating different products in different categories. One thing that I realized which is also very interesting if we're gonna go into a higher level of optimization for this is that the way the AI, the Amazon's AI built listing for the home and kitchen category is drastically, drastically different to how they built listing for the electronics category. And the way you do that, the way you see that it's because of the bullet points are in the home and kitchen category are super long. Depending on the prototype, they're way too long, and in electronics are super short and super technical. And the way they put the title, it's like super short, super technical and a lot of like compatibility words like compatible with iPhone, compatible or Bluetooth or things like that.

Speaker 2:
30:12

So then, what I said okay, I'm able obviously not myself, and I'm not gonna do that for every single prototype or category but if you want to really go deep into understanding how the system is analyzing and reading your products, do the reverse engineer for your category there. Try to create products based on prompts. You can use prompts or images, doesn't matter at this point, where you see what's the structure and the framework that Amazon's AI is using to create listings, because that's the same structure that they're using to analyze your listings. So, for example, the first thing in the electronics category is in the bullet points that I saw and I did several tries, several different attempts, different accounts. Also, the first bullet point is always the where is it from like where is it made?

Speaker 2:
31:05

the product and the dimensions, if there are any. Like you know something I don't know, tv make China and these are the dimensions and I was like, oh wait, that's interesting. So if you're selling in the electronic category, make sure that your first bullet point has that kind of information you know as part of all their things. I'm not saying that you should use solely what the AI gives to you, because I still think that they have a huge way to make it amazing.

Speaker 1:
31:33

I've tested out just for fun to see you know, based on the prompts, to have them build the listing for me, for, you know, clients that I work with and, like you said, I find it super fascinating and I think it's the smart way to look at that, at least right now, is, you know, get that data of what Amazon thinks your product is based on the prompt or the image. Use that in order to you know, craft your listing. But also you know, take into account kind of the best practices that we already have around keywords. And then also you know what you know about writing good copy and making sure that you're speaking to that end customer that the end of the day, is gonna, you know, pay money in order to buy your product.

Speaker 2:
32:15

Yeah, no, totally.

Speaker 2:
32:18

I think that's a.

Speaker 2:
32:19

That's the part where you know in the past where we used to be so obsessed with keywords and we were doing keyword stuffing but at the but but honestly, that was in like very good for the shopping experience and customers were seeing.

Speaker 2:
32:33

So we're still in that. We're in that land where you know we need to please the AI, because we definitely need the AI to know what our product is about and how people should use our product and at the same time, we need to have a somewhat good listing to convert. And I say somewhat like it needs to be amazing, it needs to convert amazingly, but at the beginning I think we prioritize and because the first person or people or system that we sell to is the AI, if you don't sell to Amazon first, if the system doesn't understand your product, then you will never sell to any customers. So you know, have that in consideration, because sometimes if you focus too much on having something that is too good for the customers, too good converting, maybe that doesn't give enough information to the system and that's it. This. This is where we need to balance that, act right and say what's good or not yeah, yeah.

Speaker 1:
33:27

So I don't know if this, you know, is kind of how you think about it, but I think about it.

Speaker 1:
33:32

You know, visibility versus conversion. So playing nicely with Amazon when it comes to keywords and the prompts and you know the other things that we're talking about here, is going to give you that visibility, whether that be. You know, what you want ideally is your organic rank, where you don't have to, you know, pay for an ad every single time. But also on the ad side as well, you know, the more aligned those things are on the ad side, the more effective your ads are going to be. So that gets you a lot more visibility. It gets you, you know, think about being at the front of the store, you know, in that kind of, you know, end cap where customers can easily find you. So that's one piece, is the visibility, and then the other piece is on the human side, is the conversions. So once you've gotten part of the customers, then you need to convince them to buy. But if they never see your product, yeah, you know you're, you're pretty much dead in the water. So 100% agree well, it's part.

Speaker 3:
34:18

It's part science part are the science part. If you're not getting the science part right, we're speaking to the algorithm. You won't even show up for those customers that are looking for it at that moment. But the other thing that we've talked about is Amazon's going to be switching more to this intent-based search. Well, I mean, if you know your consumer and you know the type of people that buy your product whether or not let's like some a closet organizer, but yours is very specific for like an RV, for example if you say that in your copy, then you're going to show up for these intention-based searches for people that are looking for a camping like.

Speaker 3:
34:50

And if your images are of this closet organizer inside of an RV closet as opposed to inside of a house closet like you're you're telling Amazon this is the person that I want to show up for. So I mean the the art part of your listing copy has always been speaking to the consumer. Well, you're going to be telling Amazon in the same time, while you're speaking to that camper who needs this closet organizer, you're telling Amazon what you want to be showing up for. When that camper goes to, amazon starts hyping and other stuff. So I mean, this art part might become more science than as this new system rolls out and this Amazon's new way of looking at things.

Speaker 2:
35:26

Yeah, and that ties to the fourth recommendation for optimization, which is the place where I spend the most amount of time is flat files, so very unsexy, but I love them. There will be an extremely important tool for optimization because it's not only going to be about copy, meaning it's not only about what the system understands of your semantic copy, but also certain functions of Cosmo are coming from specific attributes on the listing. So, for example, if you're selling a mattress and they have that example in the paper a mattress and one of the uses used for in the back end of the listing you can have that there is an attribute for that you put camping. Now the system knows that that function that they're using used for it's camping. So, leveraging the uses of Cosmo like capable of use for and the other ones are directly tied to the attributes in the back end of the listing that most times sellers completely neglect or they're like very loose, like oh well, whatever. Let me put whatever thing. We need to start being very, very intentional, because one change like this was an inflatable mattress that they were putting in the example. If the inflatable mattress they use case is for, I don't know, a guest bedroom or living room or whatever I don't know how you whatever other uses, or camping. That's a completely different segment. You're talking about a completely different audience.

Speaker 2:
37:09

So the system is also looking into that and that's probably the thing that today you can do right away to make sure that you have those things there before everything rolls out and everything starts breaking. You can still change it and you can like, iterate, many, many options. This is what we do for clients, you know, understanding what it is, putting things in the back end, testing it out for a while and then coming back and keep iterating to get to the best point, because those will impact the way the system categorizes you within the prototype. So, yeah, I think those are my four recommendations that things that people could do today, like if they can start doing it, they can start understanding this, getting more familiarized.

Speaker 2:
37:59

When is this going to get rolled out? When is Cosmo going to go out and mess up with the whole thing? I don't know. But one thing I know that is for sure going to happen, like 100% sure. That's going to happen and if you're prepared for it, then, as you said we mentioned at the beginning, you're going to get way more or in a better position than the people that are completely unaware of this change.

Speaker 1:
38:26

Well, yeah, I just want to just kind of double click on that a little bit, because you know for people for the most people it's going to take a while for them to transition over to kind of this intent-based search. However, with that said, for the early adopters, I mean, think about if you were able to essentially roll back the clock for five years on your listing optimization and on your ads and where your cost per clicks and those types of things were at at that time. What I hear out of this, or I wonder if, moving towards this intent on base search, well, it may not be a huge portion of the searches right away. It essentially becomes one of these blue oceans where there's not a ton of competition, and so it becomes a huge opportunity.

Speaker 2:
39:14

Well, john, but the thing is that the most interesting part well, I guess I said that every time I say something about the people, but the conclusion in the conclusion, they said we did AV tests, we experimented this with this new algorithm with 10% of all Amazon US traffic.

Speaker 2:
39:37

So then I went to numbers and the structures revealed a notable 0.7 increase in product sales within this segment. So I went back to numbers and I said, okay, amazon did $700 billion in 2023. 10% of that is whatever, whatever. And then the 0.7% of that is $490 million that they were able to increase just because they were targeting intentions versus just keywords. And that's the flow that they were taking customers in the navigation bar with the filters on the attributes that you have in the backend of your listing. Just with that, they already did this. So I mean, this is a no-brainer for them and that's why I'm saying it's going to get implemented. And we're talking about millions and millions of dollars that they were not getting just now with them, just with keywords, but matching also intentions. So I mean, I think it's huge. It's huge.

Speaker 1:
40:46

Yeah, and I think it's great, like you said, to be on the forefront of this because it gives you such an advantage. So one question that comes to my mind is you talked about how the logarithm interprets that image. How much of this do you think could sellers benefit from even taking their current images and essentially using that data from Amazon and how they're interpreting them in order to use AI to essentially modify or improve their images? To speak better, essentially to the Amazon AI.

Speaker 2:
41:24

Well, percentually I don't know how much better it will be, but I can tell you how bad it is when it goes wrong To the point where you start they can change your category to a point where you can start not indexing in things. That youand this is also a problem that I've seen lately is I don't know why my listing is not indexing in the keywords that I was indexing before. I'm not ranking what happened. I'm not driving as many cells as I was driving from that specific keyword. That was our number one and we kind of did the research and investigation of like what can it be? And we didn't find anything that was obvious. It's not like, oh well, you don't have the keyword anymore in your backend or you're not advertising anymore or you're sometimes restricted. We didn't find anything like that.

Speaker 2:
42:15

So this is where I was thinking like, okay, it might be that their system is not interpreting you in the way thator for that specific keyword and I know Brian also talks about how your indexation and ranking depends on the velocity that you have for specific keywords. That still happened. So this is where I'm like we are in a transition period where I don't knowI don't think I have a proper answer to your question, because in one side we're still selling keywords and on the other side we have a system analyzing our images and re-categorizing us. So I don't know when that will blend and say like, this is the system that we're using and now your images will be critical. But another cue to this was the news I don't know if you saw it in Seller Central where now they are requiring three images on your listing the main image to be like white background, lifestyle pic and technical details.

Speaker 2:
43:21

And you are like okay, before it was a guideline, before it was like you better have this, now it's a requirement. So another cue of how important your images will be, because if you don't have those three at least those three, the system wouldn't know what is your problem about. So they're taking a lot of weight into the images and I think we like as a marketplace in general and as a world and consumers and all of it, we're going to images. Now it's every time it's more images, more videos, more like visual content. So I see how that could be playing a huge role where the copy is not as important. It will always be important, but it won't be as important as the images.

Speaker 1:
44:11

Interesting. So one of the things that you said that I wanted to just spend a little bit more time on. So you talk about flat files and I feel like a lot of sellers. As soon as you say flat files, there's usually some sort of you know swear word that comes after that.

Speaker 3:
44:26

So yeah, they always roll back in their head. Yeah, yeah, so start to get a little twitchy.

Speaker 1:
44:33

And I think you're spot on in saying that. You know, I don't know if they'll ever get to the point of loving flat files, but at least to the point where they can accept flat files, or how do we get them to that point? Or what recommendations you have for sellers are listening and they're like I don't want to do flat files, but I know I have to. What are some, you know, maybe some first steps for them using flat files in order to positively impact their listings and their products?

Speaker 3:
45:06

I can actually speak for real quick before you dive into that. Vanessa, something that you actually taught me is that they aren't as scary as they look. I was scared of them just because of how they looked and all the different tabs, and it was speaking a language that I didn't speak, even though I sold on Amazon. So what I learned from Vanessa is that once you actually read, read it and look at it and what all the fields are, it's not as scary as it actually looks. So go ahead and answer the question.

Speaker 2:
45:32

Yeah, it's like for sellers, the most scary things are death taxes and flat files.

Speaker 3:
45:38

So I'm like yeah, it's okay.

Speaker 2:
45:42

As Matt said, it's actually like it's always something that's still uncomplex. It's always scary and, I'll be honest, when I started I was like this is kind of intimidating. But it's just a template and the template has basic principles basic rows, basic structure, basic like it's a. It's actually a system where you have like input something and get an output. It's in that sense, if you think about that framework of like, okay, I put this here and then it spills that out Amazing, it's not that complex Then. Then the thing that gets overwhelming probably is how powerful they are.

Speaker 2:
46:24

And the reason why people are scared is because if they change something and that something was wrong and then they upload it, then the listing will break. And it's true, it's true, that can happen to you. Actually, I did a workshop, a full day, like nine hour workshop, in Miami with a bunch of people, the computer open, doing flat files and stuff, and one of them, like when he uploaded the, the flat file remove all of his images and I'm like, okay, no worries, I'm here, we're going to fix it. It's a reality, guys. Like I'm not going to lie flat files. If you do it wrong, it can have a massive impact. So that's why people are like I prefer not to do it, right, but if you really read the instructions and I also think that's one of the one of the reasons why people don't like it either it's because you really need to be detail oriented, and most people these days don't even read nothing. Basically, we have a box where I agree to the contract, whatever, sign me up, sell my house, my life, my soul, whatever. So we don't read it, so it's just like downloading something.

Speaker 2:
47:35

Okay, and the first step, just baby steps. Right, download your category listing report. I cannot believe that to this day. Sometimes I get people and I was on a call one week ago, actually Wednesday last week massive community and I was talking about the category listing report and somebody in the chat said, like how can I download it? And I'm like, are you talking that you're being selling on Amazon for four years and you have never downloaded your category listing report? That's incredible. So what the category listing report is is the database of every single attribute that you have inputting to your listing, so titles, description, bullet points, technical details, ubc, as in all of those things, the way to download that right now. So if you want to do the baby step. Just download it and use that as a database.

Speaker 2:
48:27

You should always have your your category listing report available somewhere. There are two options, so some accounts will have it available right away when they go to inventory and then reports, and then sorry, sorry, sorry reports, then inventory reports, and then in the drop down menu and inventory reports, you see a category listing report. If you don't see it that, if you don't see it there in the in the drop down, you need to open the case with seller support and request access to the category listing report. When you request access now it will show in drop that drop down menu. So again, reports, inventory and in that dashboard drop down menu, category listing report.

Speaker 2:
49:12

What that is is your whole catalog in an spreadsheet. So you will see there every single attribute and the thing here okay, you already did it. Scary, looks horrible. Take a look at it like every single day for a week and then, on the date eight, what you're going to do is notice the attributes, meaning the columns that are empty, and that's your queue. If you have an empty column, you have a problem. Well, not a problem, but you're missing information.

Speaker 1:
49:45

I missed opportunity.

Speaker 2:
49:47

I missed opportunity and now, with this new system, those empty spaces will be critical for you to leverage. So next step is just update the things that are empty. Don't change anything, Just update the things that are empty. Some things will not be applicable for your asin and that's totally fine. You can leave it empty or you can put whatever you want. I also have tested a bunch with the flood files and I know that if something is not applicable to you and you still put information, nothing will happen. No, nothing will break. Nothing will show up in the front end of the asin, so it doesn't really matter. And then upload it and make sure when you upload it that you use partial update so you don't break anything right away, because I probably won't be there to help you, but if anything breaks, just reach out.

Speaker 2:
50:38

I always get people. It's so funny. One thing I get a lot this is sometimes I'm like I'm so proud of you guys that you are taking a step into the flood file world, but right now the flood files hide the third row and sometimes they hide some columns. By the way, that's another thing. If you download your category list in report right now, you will see on top of the letters of your columns, you will see two plus signs with a one and a two. You need to click the one with the two to open every single column possible in your file.

Speaker 2:
51:13

So a lot of people have messaged me and said like, hey, Vanessa, I don't see the columns that you're talking about and I'm like, yeah, I need to click the plus sign. So that's like amazing. I mean, you're taking a step and it's a huge thing that you're managing the flood files. Now it's just like, okay, what happens when you upload it? If you are doing partial update, you won't break anything critical and if somebody wants to learn more and they're interested and they think that flood files are the way to save their listings, which I think it is I have a bunch of content on YouTube. You can find my name, Vanessa Hong flood files, and they are like probably five or six master classes that I've done with flat files, sharing my screen and all of that. So, yeah, if they want to take action, they could do it today.

Speaker 1:
52:01

Fantastic One I think that's an amazing challenge to put out to our listeners is, if you have not downloaded your category flat file, your listing report, thank you, go do that. And if you've done it before but haven't done it in a while I think that's such a great tip is look in there to see where those blanks are at and think about what should go into those blanks, because, like you said, vanessa, those are missed opportunities when you just leave those things blank, and I think those, based on our discussion that we've had today with kind of that, how AI and cosmos is going to be becoming more important on Amazon those missed opportunities are only going to get bigger over time. So I think that's a fantastic challenge for our listeners out there. You mentioned the YouTube channel and the master classes you have there for folks that want to learn more about you and more about you know all the things that we talked about today and a lot more involving catalogs and catalog management. Where's a good place for them to find you? Oh, I think you're muted.

Speaker 2:
53:12

Sorry, linkedin is my favorite network right now, so if you want to reach out there, connect there, follow Vanessa Hong. Instagram is at Vanessa Hong and if you want to email Vanessa at onlineseller solutions on the website to the agency is onlinesellersolutionscom. You will see there. There is a bunch of things happening this month Well, sorry, next month, huge changes with the agency that I'm very, very excited. So if stay tuned, it will be really, really cool.

Speaker 1:
53:45

Fantastic, yeah, and I think, a lot of great information. Thank you so much, vanessa, for kind of you know, breaking the news, if you will, on the AI connection with Amazon and how it's going to be probably significantly changing the logarithm and how we optimize listings in the future. So I appreciate you bringing that to us and thank you so much for being on the podcast.

Speaker 2:
54:10

Thank you Thanks.

Speaker 1:
54:12

Vanessa.