AI Foundations on Amazon - Rufus In-depth

Vanessa Hung.m4a [ 00:00:00 ]Hey guys, Vanessa Hung here and today I'm at the VDSS Market Masters Think Tank in Austin and I'm going to present you something that is really fundamental for the next three or four years of selling on Amazon. Well, I think the future of e-commerce is going into AI and I don't know if you've seen anything that I have posted about the foundation of AI on Amazon and how it's shifting the platform. Today, I want to focus on Rufus, our friend here, and this is for the definitions for the ones of you that haven't seen this. Rufus is the front-end shopping assistant that Amazon gives to consumers for them to find better products through a conversational experience, yes? So then we have Cosmo, which is the algorithm that complements the A9 in order to understand the value of the product. [ 00:00:55 ] So, let's get started. So, let's get started. So, let's get started. So, let's get started. We understand and rank those products. Not only Cosmo understands the catalog and what the product is, but also the function and it understands the customer intent. So, both things in the back-end work together and we can influence Cosmo based on or by changing our data. Then the last piece of the puzzle is DSP. This is the massive shift that we're seeing in the platform. And in the next years, we are really going to see a change in the way that people shop and sell on Amazon, because it's really focusing on target audiences. But we're not going to get there. We're going to focus on how we can influence this thing here so it serves better to our purpose. [ 00:01:41 ] The first thing that I'm going to do is I'm going to see my listing. So in this example, I have one listing that we are going to optimize for Rufus, right? So in the desktop experience, if you go down the listing, you will see here. So in this section, you will see the questions that are prompted by Rufus, again, the AI personal shopper for customers to understand what your product is about. So in this case, I have, are these pillows suitable for all sleeping positions? Does these pillows retain their shape well? Does this pillow are hypoallergenic? And so on. So those are the perfect things because that's probably what your target audience is asking, right? And we want to leverage those because we want to know if our product is actually answering those questions. [ 00:02:42 ] And the way that the AI works is they use the data from the customers to understand what they're looking for, okay? So in this case, we're going to say, okay, is it hypoallergenic? Perfect. This product does not mention anything about the pillows being hypoallergenic. One thing that I know about this and this specific category is that the hypoallergenic keyword is a pesticide. So you probably won't be able to have it in your listing because you will go down unless you have the EPA and all of that. But you need to test this on your listing to see if the questions that Cosmo is prompting, you have information in your listing to answer. So if you have a question that you need to answer, you can answer those. [ 00:03:24 ] In this case, they fail because they don't have that information. And I understand because of the Amazon, you know, algorithm and restrictions, that that's not, you know, possible in that case, or it is if they have the EPA, but there's no information. So going back to or going to our other example, imagine that this is a competitor, then I'll go to the Rufus section. And I see that there is a similar question, are these pillows hypoallergenic and safe? Okay, perfect. So in this case, this guy says, or Rufus says about this product, the product description states that these pillows are hypoallergenic and made with skin-friendly microfiber. So this is extremely important. Why? Because you or what you get prompted through Rufus is what the customers are mostly asking about. [ 00:04:23 ] The product, most of the time, is about functionality, you know, will I be able to use any specific locations, functions and stuff like that, you really need to make sure that you're answering those questions. So what is that you're gonna do after you see your competition, right, you need to go to your competitors and see what are the questions that Rufus shows there, and if they're answering. So you're gonna gather this data, look at the competition, look at the questions that are also in your listing in the Rufus section of your listing, and make a list of all of the things that you see there. You don't need to worry about the answer that Rufus gives just yet. [ 00:05:05 ] It's important that you flag the ones that it's answering by the product description doesn't state whatever it is, because that's where the ones that we want to answer and control. Okay, so we're going to make a list of all of the questions that we see. And by the way, once you click in one question, Rufus will prompt you. It will prompt follow-up questions. So you want to go down into finding like what are all of the questions related to this category, or this specific prototype, because that's what the customer is asking. Okay, so you're going to make a list could be 40, 50, how many questions that you want? Okay, so you're going to make the list of questions. Yes. How many, like whatever number that you decide depends on how many competitors you have. [ 00:05:55 ] Sometimes the questions get very repetitive. So you don't need to like repeat questions, but just make the list after you have that list. You're going to go to your favorite AI, whatever that is. I personally use ChatGPT. Feel free to use whatever. And you're going to put the information of your listing. And so you're going to put the title, the description, the information that are in your images. And you're going to put the list of questions that you got through your, you know, research, finding the questions in the competition through Rufus. Then you put it in ChatGPT or whatever AI. Okay, so listing information. Then questions. And your prompt as a prompt. You're going to ask, you're going to ask the AI, is my, is the information that you see here, is the information of my listing answering the questions that I just gave you? [ 00:07:08 ] Okay. And it's important. I like it. I like to do like a scorecard. You can, you can leave it like this. And with this question, you will get the answers. But what I like to ask also is like build a scorecard on how good I'm doing to answer these questions from a one to five, you know, rating. So being five the highest rating. So what's going to happen when you put that prompt in is that you're going to get a scorecard like this with the question, with the score, and with the description on why you're not hitting your target. So for example, if the question was about the hypoallergenic, the first question, the system will read your content and it will tell you, you are one out of a five. [ 00:07:56 ] And in the description, it will say the reason why this is one out of a five. If there is no mentioning of why or if this product is hypoallergenic. Perfect. You want to know your score in all of those questions. Some of them, hopefully you will max the score, right? Like five out of five. That's amazing. That's what you want. Focus then on the one out of five. The ones that are three, two or ones, right? And recreate the content. One big thing about this shift in AI and the experience for not only the shoppers, but also the platform is that we need to give context for our product. We need to say more than just what it is. OK, so, yeah, it's a [product]. Yeah, it goes in a [bed]. [ 00:08:44 ] Perfect. But then it's like you need to create that semantic SEO and answer those questions. And one other thing, best ways to answer the questions will be also consider your images. So you can answer a lot of questions through images. For example, in the specific example of the pillow, you could have one image with, you know, three people sleeping in different positions. And with that, the system is so sophisticated. You don't even need to say it fits all sleeping size or types. With that image is so powerful that the system understands that. And gets it from you. So what you're going to use is this scorecard to build better content and do that in a recurrent basis, because next time when I go to this listing that I want to optimize, next time when I look at the answer, Rufus will say like, yeah, this probably is not hypoallergenic, right? [ 00:09:39 ] Because you stated in your content that that's not the case. Well, again, we're in a tough position because the hypoallergenic keyword is kind of tricky. But to put another example is like, is this good for camping, right? Is this a pillow that is waterproof? I don't know. Things like that. You need to answer the feature. You need to answer the question that your audience is asking. Then once you have that, because this time at this point, what you're doing is reacting to the reacting to the change of the system. You want to answer what your audience is asking. And then the other part, which is going back to the pyramid here, you have Rufus. That's what you're influencing when you change content is through Cosmo. [ 00:10:31 ] So you're teaching Cosmo new content about, yeah, my pillow is for all sleeping positions. It's best for travel. It's hypoallergenic, sensitive skin, silk or whatever. So for the last part, which I won't have enough time to get into, because it's very deep, is the DSP model. So think about DSP. DSP is the demand side platform that Amazon uses to advertise in other channels and publishers. Right. So that's what they used to publish in or advertise on The New York Times and outlets like those. But the most important part for us in this, in this particular system is the data that they have about the customers. So DSP is a massive platform for you to understand who your real target audience is. [ 00:11:22 ] Right now, with what I just showed you, we're just scratching the surface of what's there and what they're asking you. But if you want to go deeper, if you want to build your brand to the next level where you are really owning your your audience and owning that target, then you will go through DSP because that's where you find who your real audience is, who's buying from you. You can target specific demographics. You can do micro segmentation with content and all of that. Those things that will really matter in the future of the Amazon world on selling on Amazon. So if you want to learn more, follow me on social media. My favorite place is LinkedIn. So you can find me as Vanessa Hung. You can also send me an email at OnlineSellerSolutions.com. Well, we can; I can help you navigate through the changes with Cosmo and Rufus. And also, if you want training, I have a lot of training or I've been doing a lot of training around this idea and how the whole ecosystem will change, especially on the DSP. If you don't know about this tool, you should get familiarized with it because it will drastically impact the performance of your business in the near future. Thank you, guys.