When a customer types a question into ChatGPT or Perplexity, how does it decide which brand to mention and what to say?
AI is changing how consumers find and choose products. This year, shoppers are asking conversational, context-rich questions, and large language models (LLMs) are generating answers instead of search results. So how can marketers make sure their products and locations surface when there's no "results page" to optimize for?
The episode breaks down:
How AI assistants like ChatGPT, Gemini, and Claude are collapsing the marketing funnel into a single conversation
Why structured data is now your most valuable marketing asset
How "prompt inversion" changes discovery, turning broad queries into high-intent recommendations
What AI citation patterns reveal about which brands get referenced and why
Practical steps every marketer can take today to make their content, offers, and inventory AI-ready
Plus, Rebecca and Christian unpack what Walmart and Amazon's AI strategies signal for the future of commerce, and how brands can show up consistently everywhere AI relies on.
If you're a retail or marketing leader preparing for an AI-first holiday season, this episode will help you reimagine visibility for a world where discovery happens inside conversation.
Episode Links
Transcript
00:00:00.040 — 00:00:52.800 Black Friday is just around the corner. Historically, I have dreaded holiday shopping, especially for those really hard-to-please friends and family members. But this year it's different. And it's because I have a secret weapon. My secret weapon? ChatGPT. Uh, instead of generic articles and dozens of Google links, I am getting highly targeted recommendations within my budget, and I know that I am not the only one who is making this shift.
This is a really significant change for consumers and is having a really big impact on local retailers. So to help me break it down, I've invited Christian Ward to join us on the episode today, and we're going to talk about the impact of AI on holiday shopping, how Walmart and Amazon are taking wildly different approaches to thinking about this, and what brands need to do today to make sure they show up in those searches tomorrow.
00:01:00.940 — 00:04:11.890
Hi, Christian. Hello, Rebecca. How are you? Great. Thank you so much for joining. There is so much that's happening. Uh, Black Friday is just a couple of weeks away. Getting ready for the holiday season. Um, and my guess is that a record number of people are going to be turning to LLMS to help them do their holiday shopping. I definitely will be, especially for my dad.
I'd be curious to know, why is this such a big shift, and what impact do you think it's going to have this year? Well, first off, um, it's a great question because the adoption rate of AI compared to last year has more than doubled. In almost every survey you see, I think ChatGPT has close to 800 million users of just their app.
It doesn't even count all the other LLMs that are out there. And so I think what you're seeing is, is this sort of perfect wave of adoption has built this whole year, and now you have people that are much more confident that there's a value or a fair value exchange in using these tools to help them identify perhaps the perfect gift or even just streamline their process in finding that gift.
I can't wait. It's going to be so much better. It is happening really fast, and I've noticed that retailers are responding in very different ways. I think we had a study we saw about Walmart, saying like 20% of its referral traffic is coming from AI, whereas Amazon just filed a lawsuit against Perplexity to block all those automated shopping agents.
So why do you think they're taking such radically different approaches? And what do you think that's telling us about where the landscape is going? Yeah, it's actually fascinating. I think this is probably the best way to think about it is it's a bit of A Tale of Two Cities right now, which is you have a lot of retailers that have spent an enormous amount of time and money to build out their presence.
So it's an Amazon, clearly a dominant force in almost all shopping, let alone holiday shopping. But Amazon's taking a very different approach, saying, listen, we don't want these bots crawling our site. We don't want to index our data. We really want to maintain that direct, one-on-one relationship with the consumer.
And to be fair, Amazon owns a pretty big chunk of Anthropic, so they've got some really good models that they can use to try and optimize for that. On the flip side, you have Walmart, which I just want to remind everybody I'm very proud of Jet.com being a Hoboken, New Jersey town, because I lived there for so many years.
It was like a hometown hero for us. Jet.com got bought by Walmart in 2016 for roughly three point something billion dollars, and that became their online platform for all things product and sales. And it's amazing that Walmart comes out and says 20% of all of our referrals are coming from ChatGPT.
Why don't we do a deal with them? We didn't have to go to Walmart.com. You can literally learn the product and tell Chat GPT to buy it for you. These are two very different approaches, and I think we're going to see that same sort of dichotomous world play out for a while between those who are fighting against losing that one-on-one direct, with those that are absolutely okay with it and are taking a much more aggressive approach of being everywhere.
It's so interesting. It's almost as if it reminds me of the businesses and the dot-com boom. That said, they were never going to go online.
00:04:13.930 — 00:14:54.200
They usually disappeared. Um, it's been very interesting. It was a very adorable take. They're like, oh, this internet thing. It's a fad. Um, it's not a bad AI, it's not a fad. We're pretty sure, um, there was a Financial Times chart that showed that in two years, there are as many people just using ChatGPT as it took 14 years for people to use the internet, the entire internet.
So it's sort of disturbing how fast it's being adopted. I do not think the right play is to bet against it. I agree, I totally agree, and it's interesting because if I imagine someone skipping my website entirely and still purchasing my product, what it seems like is that the traditional funnel, the how I get found, the awareness, the consideration of valuation evaluation conversion.
It's totally collapsed, and it's all moving into these LLMS. So if ChatGPT or some other platform is handling all of that on my behalf, um, my marketing strategy needs to change a lot. So I'd like to understand, like, how should we be thinking about changing our marketing strategies? Yeah. Look, again, I think platforms that are sort of aggregators.
So Amazon's classic aggregator, Walmart's also technically an aggregator. They just take very different approaches to it. Anything where the AI almost has a more acceptable or usable interface, you're going to see that top of the funnel and the middle of the funnel move into that sort of chat dialog.
Um, and while I know every brand that's our clients or our prospects, I love you all, but can I tell you something? You're probably not going to have me as a consumer choosing to have personal relationships with 500 brands' eyes. I'm going to have one AI that talks to your 500 brands like that. We already know this academically; this research has already been done.
And so if that process is where it's going. Then, for most people, they have to reconsider that their website is now a conveyor belt of structured data for these large language models. Unless again, you're going to take Amazon's approach and fight it and build your own now. Also, I don't want to count them out.
They have 600 million Alexa Echo Dots all over the world. They have a very good path to maintaining that direct funnel, but if you don't have that, it's much more likely that you need to look at your website as still critical. But it's really about the criticality is making sure that your information, your content, is everywhere and structured so that those large language models know to consider you in that funnel process.
Absolutely. And one of the thing that occurs to me when I go into ChatGPT to search for gifts for my dad, who is the hardest person in my life to shop for. And what I notice is that it asks me questions back, like when I go to Google and I say gifts for dad, I'm going to get like just a bunch of articles about the best things for him, tools and stuff and all of that.
But ChatGPT asked me questions back, and then it came back with a really great list. So as a consumer, that's great as a marketer. I imagine the number of queries that I am relevant for has become dramatically smaller. Hmm. Yeah. So, um, you know, what kind of impact do you think that's going to have?
Well, it's a great question. So the phenomenon you're describing, we call prompt inversion. And basically the idea here is, I might ask, hey, I'm looking for, you know, a gift for my mother-in-law or for my dad, as you said. And the reality is, when you do that, um, the AI is smart enough to go, well, that's not really enough information.
Why don't I ask you a few questions, and we'll get a little further? Like, what do you want to spend? Um, how, you know, how fast do you need the item? Uh, there are all these elements. So the example I typically use is if I say backpack in Google search classic search, I'm going to get 9700 million pages of backpacks.
And now that's not really helpful. What we're realizing is that it was amazing in 2006. It's not really amazing now. Now the AI goes well. Wait a second, is this a backpack for like work for laptops, or is this like camping? What are we talking about here? But immediately when it does that those half of the queries where you might have shown up for back, they're gone.
Because if you're right now, you're talking about some form of a camping or hiking backpack. Two more questions and you're pretty much at the bottom of the funnel already, where there are only three backpacks that kind of meet what I'm looking for. So I think that's the realization that consumers are going through, which is, wow, this thing streamlines this process.
Instead of having 92 tabs open in my browser because I'm looking at so many backpacks, it kind of gets you there really quickly. And I think that is a very big opportunity. But to your point, it means that we almost shouldn't look at it as, hey, I'm not getting the impressions I used to get. Well, we really need to focus on is, yes.
But the people that are landing on my site are really the they're the people that were going to get there anyway. It just took them three clicks at the to zero pop and not 728. And so that's the beauty of this is it's it's more efficient for everyone. But it does mean that as marketers, we're giving up that impression-based KPI existence.
And we're moving more to the world of, hey, can you help the people use my data to come to me? So on that note, I think it just becomes more critical than it's ever been to ensure that we are showing up in those relevant searches, right? Those queries. And so, um, to do that, we need to be giving structured, trusted information to the LLMS like we talked about on our last episode.
And so I want to pivot to, to revisit that topic a little bit, because there was a fascinating new study that came out that really validated the data that we published a couple of weeks ago. So can you talk a little bit about that? Sure. So there was a study done out of Europe. I think it was actually the Max Planck Institute.
And this it the article I actually saw was in Ars Technica. So it wasn't even necessarily something, it was more just a classic sort of technology-based Analysis, and they came up with this study. I think it was on a million citations. And what they found was a lot of the sources of the data for things like shopping queries or brand queries, anything in sort of that process.
But it came out with this and said it basically like less popular sites were being cited a lot more than the very large popular sites. And there's probably a lot of reasons for this, but it directly supports the study. We launched two and a half weeks ago that said, on a study of 7 million citations, we found that all the old directories, small aggregators, all these platforms where you're bringing in content information, they were really critical.
They were showing up a lot. Whereas a lot of the brand websites, those were also doing really well. Um, and so it gives you this idea of the kind of got to get all of this right. You cannot mess this up and sort of be, you know, frugal with your data. Um, because remember, these models, they're all probability-based.
The more I see your product with this offer in this inventory level at this store. The more I see that as a large language model, the more confident I am that that's the right data. And so it's a very simple formula. But that study was great. It literally was just a few weeks later and it said pretty much exactly what we said.
Well, it's really interesting. I love what you're saying about dumping frugal with your data and consistency everywhere to tie into the probability model, but why do you think they're citing these less popular sources in the first place? Yeah, it's funny because, um, they call it less popular. I actually would say you mean the really well-structured ones that we all used to live our whole lives using six, ten years ago.
So things like MapQuest and show me local and yellow pages. I don't know if you all remember, I used to like, print my MapQuest and put it next to me while I was driving. Like, we used to live on these platforms. Now, what people don't realize is there's probably both a technical reason why these are cited so much.
And there's probably also a legal or business reason why, which is if I start pulling Amazon's data. As Perplexity, I might be inviting a lawsuit if Amazon doesn't want me doing that. And that just happened. You might also have a situation where some of these longer, less popular sites that have been around a long time, they have enormous authority, but on top of the authority that they've been there so long, they structured the data on those websites.
There are some of the best in the world, like literally the facts and the artifacts you're looking for. They're all right there. And so what we would say to most of our clients is the more you can sort of piece together all of those, the better, the more likely that you're going to really benefit from them.
But it's I think there's both business reasons and technical reasons for why those sites are doing so well in the AI citation game. So as a marketer, what should I be doing today to increase the chances that I'm going to show up in a query for a holiday search? Yeah, well, there's a couple of things I don't think the world quite has grasped how powerful the large language models are on also things like your products, good services, That's sort of table stakes, but also your offers, your involvement in the community.
Look, it's never been easier for me to literally say to my large language model, I want to buy a great gift for my mom. But also remember, I really love supporting our local town. Can you look for businesses around me that have some cool offers? They can literally say that. And if you have a page with the offers and what you do and everything else, you're going to win.
And people have to understand that means people get way in the weeds on what makes you unique as a business really, really quickly. And so what we would typically say is all of the classic information where you are, what you have available, what your products are, what your services are. But go deeper. Start thinking about all the structured knowledge for why someone should choose your business in their holiday shopping, and the LLMS or the the the AI assistants are going to reward you for that.
The other thing that I think is going to start happening is as the AI asks you questions, it's going to ask you questions that are more effective if if they are some of these deeper meaning questions. And so you're going to see a lot more of that sort of engagement. So the more data you can put out there on your own site and through platforms that help like Yext, but push the data into all these other areas, we just want that data to be exactly the same.
So whatever you have on your site, as much that we can put out there exactly the same, that's where you get that probabilistic benefit which we we uncovered in our research. Amazing. Oh, I love the actionable takeaways. Um, okay, we've got some work to do. I think we'll wrap it up for this week. This is fantastic Thank you so much, Christian. Um, we will see you again next week for our next episode of The Visibility Brief. If you like what you heard, subscribe, share it with your team, or head to Yext.com to learn more about how how we help brands stay visible in an AI first world. Thank you so much for listening. We'll see you next time.
