What Brands Need to Know About Sparky, AEO, and GEO

The rapid shift toward AI search is causing massive anxiety for brands expecting optimal performance without a clear roadmap. With tools like Walmart's Sparky rolling out to the masses, figuring out how to balance traditional SEO with conversational commerce is critical to maintaining brand visibility and market share. Blake Taylor of AdFury.ai returns to break down the mechanics of Sparky and how brands can actually win in this evolving digital retail landscape.

We get into the exact mechanics of how agentic shopping operates within the Walmart ecosystem. The discussion covers the critical differences between open conversational queries and product-specific questions, the enduring weight of content quality scores, and why the hidden shelf description field is an untapped resource for optimization. Blake shares a crucial realization that Sparky isn't replacing the search bar just yet, but rather acting as a translation layer that relies heavily on traditional search functionality behind the scenes.

The frustrating reality of optimizing for AI is that there is no magic button or one-size-fits-all playbook to bypass the hard work. It requires a tedious commitment to the fundamentals, from filling out back-end attributes to rigorous trial and error on the platform itself. Viewers will walk away with a clear understanding of how to establish their own benchmarks, track personalization impacts, and structure their product pages to build trust with both the algorithm and the shopper.

If you care about digital commerce, organic ranking, and navigating the future of AI in retail, you’ll get a lot from this. Please subscribe and share this episode with your team to help us keep bringing you these deep industry insights. What specific product detail page attribute are you prioritizing first to adapt to these new shopping behaviors?

Transcript

SPEAKER_00 0:08

What's up, party people? BV here. Welcome to another episode of Retail Media Vibes, a doing business in Bentonville podcast. We are recording live at Podcast Video Studios here in Rogers, Arkansas. So today's episode is going to be a little bit different. It's going to be a little bit more focused and going to be centered around agentic shopping and AEO. So my guest today is actually a returning guest. And our guest today is Blake Taylor of AdFury.ai. So welcome back to the show. How are you? Good, man. Sad to be back. Yeah. So, Blake, you know, um, we'll get into, you know, what you've been up to since we last had you on the show. But, you know, in that, in that workshop, you know, you focused in on Sparky, and we'll get into a little bit more about what Sparky is. Um, and then like how it works today, right? So these things are evolving, how shopping is evolving, how AI is playing a role in that shopping experience. And, you know, I think what our intent was in that workshop was to bring some of these topics to, you know, brands specifically and some other uh attendees in that audience and help them understand, okay, this is the landscape of where AI is and it comes to the retail environment. Um and so as we talk through this, you know, I'm you know, you're going to be talking a little bit more about the Sparky stuff. I'll talk a little bit more about the content. And so, you know, it was an interesting workshop, nonetheless. And it was uh, I think super valuable to the audience. I'm sure there were a couple of you know key moments or takeaways that you might have identified in that in that workshop. What was something from that workshop that you thought you thought was, you know, a good takeaway or something that stuck with you?

SPEAKER_01 1:59

Yeah, I mean, I think the the biggest thing that stands out to me is just how fluid this situation still is and like the big claims that are sitting out there that I think almost bring a little bit of like a fear factor in a little bit of like, oh, search is gonna change right at any moment now. Yep. Um and whenever you really start to kind of peel back a little bit, um, there's some really cool things that are happening, but we may be a few steps away from like fully changing how people find products in in the world of digital commerce. Right. Um, so I think that was very, very evident. And I think as we kind of unpack uh like Sparky, especially today, like we'll see a lot of the same fundamentals that especially brands have to follow uh that are still present today. And it's not necessarily an entirely new playbook, uh, but the the output to shoppers is starting to evolve and change. And so it'll be interesting to see how that really, I guess, like gets digested by customers over the next couple of years and how quick it really changes shopper behavior and a lot of those downstream pieces. So I think that was a big piece. Um, and I think just the the different places that AI is present, like it's it's kind of everywhere. Like you've got a lot of these kind of big players of Anthropic and Google and Open AI, and they're all kind of doing their own things, but also trying to play the copycat game and like their own styles. Yeah. Um, so it's very interesting to see how everyone is approaching it similarly but differently at the same time. Uh, and then you've obviously got Walmart and Amazon, the two major retailer powers, right? And they're trying to figure out do they in-house and build everything with their own kind of tech and capabilities, like an Amazon, and building that kind of customer trust and value chain within their own walls. And you got a lot of Walmarts that are trying to figure out how do I be the best partner and how do I leverage these other companies to reach my shopper base where they're at. Um, so it's very interesting. Like there's no one size fits all answer to this digital commerce AI answer. And I think that's the hard part for brands is there isn't just a click button fix problem here.

SPEAKER_00 3:58

What I took away from it is it feels like there's a lot of anxiety in the space, right? Like there's so many demands because you know, AI, agentic shopping is in the trades all the time. Everybody's talking about it. There's this development, that development, a lot to keep up with, a lot of change in that space.

SPEAKER_01 4:14

Yeah.

SPEAKER_00 4:14

Yet brands are expecting optimal performance regardless of what state that implementation may be with a specific retailer. How do I win? How do I make sure my product shows up? And to be to be honest, like there's no answer, complete answer for that right now, right? There are some clues, there's some hints, and obviously we'll get into that as well.

SPEAKER_01 4:32

But they're a topic head questions. I mean, like every single session had uh a full set of questions afterwards, and like like exactly like you said, it's not uh I wasn't crying, but it is a lot of people that are just they're curious and they're they have a lot of pressure on to try to figure this out and be the first brand to figure it out. Yeah, absolutely. Um I hold the title of VP of ops at Ad Fury uh currently, uh, but spent the last couple years uh at Flywheel uh and what was previously White Spider in this kind of Bentville community, um, and really just trying to become the digital uh organic expert on Walmart.com uh is kind of where I would say it started. Obviously, that dovetails into many things, and you can't do organic well without a good paid strategy. And so it's really all the digital commerce side altogether, but really like trying to understand that organic algorithm and what makes it tick and how do brands, specifically items, uh, optimize to take advantage of the algorithm and win in the algorithm. Um, so whenever we're starting to talk about AI and Sparky and things like that, and it being the next version of search, that's obviously very intriguing to someone like myself who has tried to understand that algorithm and hearing that maybe it's changing and maybe that there's a new style of shopping that is about to occur uh or kind of inject itself into the the world, it's very intriguing. And so the the opportunity to kind of dig into Sparky and start to understand what's going on there uh is exciting, but a little bit daunting at times because as with a lot of things, Walmart, they're great at kind of putting it out there, but maybe not giving you all the information to fully understand it. Uh, there's no playbook necessarily for this. It's a lot of trial and error and sitting on the couch at night and right asking questions and plugging things in and seeing what happens and documenting and trying again and seeing if there's any differences. And so it's a lot of just uh, I don't know, back room trying to figure it out.

SPEAKER_00 6:21

And to be fair for Walmart, right? This still is a technology as it's evolved, right? So this is not a very mature technology. So how their approach and how they implement AI in that shopping experience is definitely going to be evolved, is going to evolve. It's not going to be perfect from day one. And so it's it's exciting to see it roll out.

SPEAKER_01 6:41

And I think the word that comes to mind is potential.

SPEAKER_00 6:44

Yeah, yeah, yeah, for sure. So we've thrown around the word sparky like a million times already in this conversation, right? So, you know, maybe, you know, maybe somebody's playing a drinking game out there and taking a shot every time we say we say sparky, but we said it quite a few times. Yeah. But let's just let's just define what is sparky, just what it is. Explain that to the audience.

SPEAKER_01 7:07

Yeah. Uh the best way to describe it is it's their shopper-facing agent uh that was traditionally on the app. Uh I think it actually just rolled out on desktop over the last couple of weeks uh for those kind of signed-in users on desktop, but it is Walmart's shopping assistant, more or less. So you can find your products easier. It will tie into your account and understand how you purchase. Uh, you can ask questions of products, of queries, of kind of all these different spaces on Walmart.com that we can dig into a little bit more today. Yeah. But it is their shopping assistant that is to help ride alongside shoppers and make their shopping experience easier.

SPEAKER_00 7:42

Okay.

SPEAKER_01 7:42

So, how would you describe that experience, that Sparky experience for shoppers? I think when you sit down with the average person, you kind of ask them like what they're seeing out of Sparky, and maybe even myself. It's such a new space that I don't think that there is a definition of like this is what I expected to see out of this. Right. Um, what we'll kind of dig into here is like it surfaces maybe a lot more products that I wouldn't have expected to surface. If I if I search for the best multivitamins for 30-year-old men or something like that, or men in their 30s, um, you're gonna potentially get a different set of items than you would if you just searched multivitamins and filter to mention. And um, and there's reasons for that. And there's back-end kind of processes that are taking place that kind of return those different results. Um, and so I think the the recall of items is still something that they're trying to figure out. And how do they find the right mix? Is it to just return the top five best-selling items based on one keyword in the search query? Like, is it the job of the of Sparky to understand the query and go and find new products to surface to the shopper? No, I think it's probably on the roadmap.

SPEAKER_00 8:49

Yeah, ultimately, I think you know, Sparky is there to help Walmart deliver on the conversational commerce side of shopping. You know, I think there's this big bet that's being paid to this is the new way of shopping.

SPEAKER_01 9:04

Yeah. I mean, the the big bet there is that shoppers shift the way that they look for products. And I'm having brunch this weekend with six people. Help me build out the menu for that or help me build my basket for all that. And it's gonna help return all the products needed to shorten that shopper journey so that you don't have to go one by one. It pulls it all into this experience uh in this moment that's being created.

SPEAKER_00 9:27

Yeah, it's it's yeah, it is definitely interesting on where they they see that you know that direction going for specific, you know, shopper uh shopper occasions.

SPEAKER_01 9:39

I think the kind of average user today probably isn't in that help me plan my brunch for this weekend category. It's what's the best ice cream flavor, what's the best vanilla ice cream flavor? Uh, and trying to understand categories better or understand products better, right? I I think that um the kind of different phase or not phases, the different kind of areas that Sparky works in. One of those is directly on PDP. So whenever you land on a product page, you can ask questions about that product. So if the content isn't clear enough to you, is this a gluten-free item? Is this fill in the blank, whatever question you have about the product, you can now ask that directly of Sparky and they can go and search um the product information that maybe you don't even see directly on the PDP, some of those behind-the-scenes attributes uh and help give you confidence in what product you're buying. Right. So I think that there's potential for maybe that to even take off more of kind of the asking questions of Sparky and rather than like product finding and product searching, maybe a little bit more of like questionnaire of like help me find the best of this, or does this product fit this specification?

SPEAKER_00 10:45

Yeah, I think to that point, you know, um, you know, so if we look at the ways that you can engage with in Spark with Sparky or in search, you know, within you know the dot-com experience, like you can have an open conversation, right? You can provide search questions and then you can also ask product level questions. Yeah in your experience with Sparky, how do those different how do how do those differentiate themselves? How are they different?

SPEAKER_01 11:13

Yeah, I mean, they they appear in different places within Sparky. So when you're on the homepage of Walmart.com and you pop open the little Sparky button, you're probably just gonna start with just a generic conversation of um planning a meal, or they're even gonna have those kind of preset kind of pills set up in there. So if there's a moment that's coming up, like Father's Day, there may be plan a Father's Day lunch or something like that, already preset into your uh kind of Sparky interface, or reorder my essentials or previously ordered items. And so it's gonna have that understanding of you as a shopper. Um, or you can just open up the conversation wherever you want to go with Sparky at that point. Um, if you are already on a search grid, if you search for eggs and now you're on the search grid and you're looking at 40 different kinds of eggs that you can purchase and different types and different colors and different free range versus cage-free versus whatever all those different types of eggs are. And you're trying to figure out what am I looking at here? What is the difference between cage-free and free-range chicken eggs? Um, you could ask that question of the keyword results and understanding the space that you're looking at and trying to understand it better. Um, Sparky will help go and source that information and will be that kind of external engine. We'll go even search brand sites and we'll look external to the walls of Walmart to return that answer to you. Yep. Um, then on the PDP itself, like we were talking about earlier, you can ask those kind of questions about products that maybe isn't clear from the content per se. Is this, does this fit this model of vehicle? Does this uh is this product gluten-free? Like it's gonna be able to search kind of the back end attributes that may be tough for you to find just on the PDP. We know the PDPs are busy. There's a lot of information that's being thrown at you. There's a lot of similar products and things that while we're trying to help connect that shopper journey to where maybe you can't find the exact piece of information you need.

SPEAKER_00 12:57

You actually, even, you know, in our research for this uh for this workshop, you did some experimentation and you had kind of an interesting experience with yogurt. Uh was it creamer, like the creamer? Oh, it was the creamer, sorry. It's Bonnie Creamer, I think. So in your research, you were, I think you were doing some testing, and I think you found some interesting results when you were looking for creamer.

SPEAKER_01 13:17

Yeah, yeah. It was uh kind of along the lines of like personalization and trying to understand how personalization plays into the results. Um and I tell you, it's somewhat inconclusive. Like I think that personalization does matter to uh to Sparky and to AI tools. Um it's hard to like dial it in exactly. Whenever our this exact example was trying to understand uh the best vanilla coffee creamers. Yeah. Um, and knowing that my wife often buys Chibani coffee creamers. That seems to be the sort of Chabonnie came. Exactly. Uh that seems to be the brand that she's had a preference towards recently. Um, I kind of expected it to be surfaced within that consideration set as um trying to understand like what drives the term the best. Is it the best ratings of reviews? Is it the best for me? Is it the best for fill in the blank? Um, and it of the five products that returned, none of them were the the Chibani item or brand that I would have expected to see right off the bat, as that's one of the previously purchased brands that we have. So that was kind of interesting. Um, I wouldn't isolate it to that one example to say like there is no personalization in here. Um, but I think I understand how personalization is folded in more um than I did in the past.

SPEAKER_00 14:32

Yeah, I think I love you know how you brought up the best. That always it always dumbfounds me that when I ask AI for what is the best of something, it never asks me the question, well, what do you think is the best? It always comes back with something, right? It's it it's crazy. Opinionated. What can so like there's obviously a transition that has to take place between using you know traditional keywords now to this conversational place. Where do you see like the keyword methodology influencing this conversational commerce and this chat bot? Where how do you see that working?

SPEAKER_01 15:13

Yeah, I think that was probably the biggest finding of kind of sitting down and really starting to dive into Sparking to understand how it works was we were kind of told that this was I'm not gonna say we were told there's been claims out there that like this is going to replace the search bar, like this is the future of shopping. That's a big claim, and that that that feels daunting in a lot of ways.

SPEAKER_00 15:33

I do think they're gonna force it into the search bar at some point.

SPEAKER_01 15:35

Yeah, I think it can absolutely be part of it. Um and so that kind of claim really, I think, starts to like can just make your head spin a little bit of like, oh my gosh, we're not gonna just like ask for a product anymore, like it's gonna be more conversational, or they're gonna try to force it to be there, which who knows if that'll fully happen. But all that to say, like, is Sparky the future? Is using a chat bot to search for products the future? Maybe. And the way that Walmart's using it was very kind of uh interesting to me. Of it isn't necessarily taking what you search in Sparky and generating a totally unique set of items to come back to you. It is kind of acting more as this like information router in a way to take what you asked it, build a keyword search out of it, and then it goes and searches that keyword kind of behind the scenes and will return products from that keyword search. So the search functionality that we all know and love is still present. It's still powering a lot of it. It is still the back end workhorse, I think, of this Sparky tool today. Um, Sparky is really the translation layer of all these terms that you use to ask a question. How do I dial that down into an actual search term that I can go and search and return products with? This is different, though, right? It's not searching for eggs anymore. It's searching for more specific keywords of uh eggs for brunch for six people. And it's building out those keywords that people would never really search on their own. And so you're gonna sometimes get some funky results as you get into those kind of longer tail keywords on Walmart.com that may have different kind of algorithmic structures than those top most volume keywords. No. Um, and so you're gonna sometimes get some funky results back in Sparky because it's using that search function, using an algorithmic process that is not necessarily for the most uh for those high volume keywords. You're gonna get more into those kind of long tail keywords that are unique, um, and you're gonna have a different experience with what items are returned. So that multivitamins for men 30 plus, like how you write that question, can then dictate how that search term gets created and what items get returned, which could be very different than just going to multivitamins and filtering the men and 30 years old target.

SPEAKER_00 17:53

So, what it sounds like to me, you're saying, is brands should not abandon their traditional search or SEO strategies right now. Yeah.

SPEAKER_01 18:05

Yeah, I think that's a great way to kind of simplify it and summarize it up. Right. SEO is not dead. Um, and we're not graduating to other areas of AEO and GEO. Like those are important and they're layering in and they they need to be thought of kind of in accompaniment with uh with the SEO component, but SEO is still king, especially at Walmart. Like it still carries so much weight. Um again, this is one specific way that Sparky works. I'm not gonna tell you that every single product finding method in Sparky is exclusive to this. You're gonna find event pages and you're gonna find more curated pages um that are using more of like Walmart's browse shelves and some of the category pages that a lot of like merchants and and dot com teams kind of set up and at Walmart. Um, and those assortments are are curated and then kind of fed into Sparky as well. So it's less about um typing in a query and then finding the results. You'll kind of notice the difference in like the page structure. If you ask a question and you kind of get a list of five products back, which seems to be kind of the standard response for Sparky, that's going to be using that search functionality to go and find a couple products and return them back to you. If you search for a brunch plan and it puts you into this whole like category experience with these like subcategories at the top, and you're kind of limited on the item sets, um, but you're seeing probably 20 items at that point, you're in a totally different experience that isn't necessarily using that search. And that is more so identifying moments and using existing category pages and and uh shelf pages that are already curated and created in the Walmart system to kind of build the assortment.

SPEAKER_00 19:34

Are are any of those results sponsored at this time? Um yes. Or or driven by sponsored. Um so if it's pulling back, let's just say it's pulling the top five.

SPEAKER_01 19:44

Yeah.

SPEAKER_00 19:45

Are you know, is one in five sponsored and two through.

SPEAKER_01 19:48

At this point, it's not as consistent as like search, where you're pretty much always gonna see 10 to 12 different sponsored products in very consistent positions. Right. It's not that saturated at this point. Um, Walmart Connect is absolutely starting to kind of test it and try to figure out how to do this in the best way. Um, whatever you have so few products that like the trust of what products are being surfaced is paramount. Um, and so it's a tough piece to try to find, I think it's tough to find the right balance of how often you serve ads.

SPEAKER_00 20:19

So you're you're hinting, hinting at, at least from my perspective, a bit on relevancy, right? Like I think adoption will tick up if people are getting positive results.

SPEAKER_01 20:34

Yeah.

SPEAKER_00 20:34

Those positive results are from relevant results, right? No matter if you're asking you know a keyword-based question or an occasion-based question, if the results are relevant to you, then you're probably gonna come back and use it because it's going to give you trust, right? And so the whole thing, and people have heard me say this a million times, but you know, the whole equation for AI adoption is trust. The more that a shopper, in this case, shopping, is using AI and they're getting back the results, that is the correct results for them, they're gonna have more trust in in the output, and so they're gonna spend more time on the input, of course. Yep. So, how from your perspective, how is relevancy gained within the Walmart ecosystem with Sparky? Like, how does it look at how I mean it was to your peanut butter and jelly example without the sponsored part of it, but just like if I'm looking for, you know, you know, peanut butter or I'm looking for what to buy to pack school lunches, you know, and it's returning results. How does it determine relevant results to put into the chat?

SPEAKER_01 21:50

Yeah, I think it it's it's still that search methodology, right? Um for a lot of queries. Again, like it's so hard to like nail this down to like one specific. Yeah, yeah, yeah. Like uh, I think example, because if you search for school lunches or I need like help me plan school lunches for the next week is your Aquarian Sparky, you may go into an experience page that's all about school lunches that Walmart has already built, and that curated assortment may already be there. And whether your item is in that curated assortment or not is more so about are you in that category page? Are you in that experience that Walmart merchandiser have already made? Internally, Walmart is using that as a source of truth.

SPEAKER_00 22:31

For the category pages. Yeah, yeah, yeah. I think so. A curated page that speaks to a specific result. Because, you know, AI like OpenAI, ChatGPT, you know, Gemini, Anthropic, uh, Claude, etc., you know, they're using, you know, brand sites and they're going out and they're they're they're trying to find the most, you know, most trustworthy uh data sources, best context uh that they use in order to provide, you know, provide results, right? As well. Those are sources that they go out and they s and they they pull in that that information, right? Yep. Walmart, and we'll talk about whether or not Walmart does that as well, goes out to brand sites. We'll talk about that in a second. But I would assume that the first place it would look at for relevancy is within its own walls, right? So, you know, PDP, category pages, brand shops, I would say, right, is as well, you know, look at your internal data set before looking externally.

SPEAKER_01 23:31

Yeah, absolutely. And again, I think it even it boils down to if your content on your PDP is kind of the first stop. And and I think Walmart is drilling this really hard of it's the it's the dirty work that has always kind of been pushed from a content quality score lens. It's now an AI lens as well, in a Sparky lens as well. It's you've got to have your back-end attributes filled out, you've got to have your title description key features in a good spot. Um, Rich Media plays a piece in this. Um and so all of these kind of just general PDP health and and cleanliness activities that have been pushed really hard by your merchants over the last couple of years, and you gotta be at that 95% content quality score mark. Like there was a reason for that. And it's getting ready for this kind of AI shift where all of those fields are considered to help give the agent Sparky confidence in the product. Um, and so it still is those fundamentals, right? I think if you go talk to anybody at Walmart, there's not some brand new shiny field or process out there that you have to go through to become optimized for AI. It's the same processes because again, whenever you go back to it and you're asking those questions, and Walmart's gonna return those five products to you in the Sparky engine, um, it's using the search functionality. So the same things that we were talking about to help your search optimization, help your organic optimization, still valid, still extremely uh fundamental to winning in search so that you can win in Sparky. Yeah.

SPEAKER_00 25:04

So from your research and and testing experimentation, was there anything that you saw that made you think that Sparky is also using context from external sources in order to drive relevant results?

SPEAKER_01 25:20

Yeah, I would tell you it's probably less at this point to drive the initial results that you see. Okay. And more so when you start asking questions of those products and of those results, it starts to then look externally to help answer those questions. Right. Um, the product finding process, Walmart keeps that pretty well internal. And apologies if I jump ahead with this answer, but like as you look at their partnerships with uh OpenAI is probably the best example at this point. They kind of launched this partnership with OpenAI. And the results that they got in this app that they built or this partnership of how people were purchasing within OpenAI products on Walmart, they saw a much lower conversion rate on that than what they saw on their own site. Yeah. And so that that product finding method that OpenAI was using was not yielding the results that Walmart would want. So, what do they do? They kind of paused that and then they kind of relaunched it using Sparky as the foundation of that application within uh OpenAI and within ChatGPT. So they believe that their product finding method, which is kind of using their own catalog and their own data, services better results and yields more purchases than maybe the methodology that an open AI was using, which probably is more sourcing out like brand information to build the initial product set.

SPEAKER_00 26:36

Yeah, I would I would also say it comes down to mindset. If you are on Walmart.com and you are engaging with Sparky or the search bar, your mindset is to shop on Walmart.com and to purchase your Walmart.com. And you're on on Chat GPT and you're you're probably more a little bit more upper funnel, like you're kind of still in that discovery process most likely. I think whether or not you're shopping Walmart or not is a little bit of a crapshoot, right? It's a you know a little bit more of a 50-50 proposition. Like maybe you you maybe you don't shop at Walmart, maybe you prefer another retailer instead, and so you're not going to go engage at the Walmart side. So I think to a certain degree, some of that would have been expected just because of mindset of shopper and the experience that they were engaging in. Yeah, that makes that makes total sense. So as we think about optimizing for AEO, geo, sparky, etc., you know, the the main source of truth is going to be the PDP, right? And so now you're engaging in a different way through natural language, yeah. Right. What are some things that you've seen as far as what changes should occur on a PDP that help with that natural language type? And I and we've already established that currently it's yeah, it you know, that those those are queries are being translated to keywords for for results, but I've got to believe that there's still some of that natural language type of processing that has to take place and have some sort of alignment within the content of the PDP to give a result that is in line with what an expectation would be.

SPEAKER_01 28:17

Yeah. I mean, we sit here and talk about this almost every single day of the different content demands that are now coming in for brands to have to meet with such limited space to be able to do that. Right. You've got your Walmart retailer requirements uh and style gap requirements, you've got uh your keywords and your kind of SEO element and make sure that those are in your titles. You got AEO and GEO now coming in and those kind of longer descriptive kind of pieces that bring authority and answer those questions. You got seasonal components, you've got uh the list can go on. I'm sure we can sit here and probably name three or four other pieces that are like vying for space within your content, uh, and there's just there's no room for it. And so I think it really does come down to like a prioritization and kind of understanding what each content field really brings to the puzzle. Uh, and maybe I'll even say it brings to different puzzles. Right. Um, when I think about the title field on Walmart, like that is that is sacred real estate as you get like 70 characters to put in there. You can't just start plugging in these long tail kind of value pieces that would appeal to AEO and GEO and answering all of the shoppers' questions in that space. So you have to be very, very direct with that. And and I would direct people to say, um you want to focus in on keywords, you want to focus in on SEO with your title field, knowing that the sparkies of the world and that that engine is still using search to product find and deliver results back. Um, knowing that the search engine is really looking for keywords within your title. I think understanding that SEO and keywords, real estate within the title is sacred, and that is the priority there. The AEO and Geo element, the seasonality, uh those start to layer in a little bit more, I would say, even in like the description and key features. While not always the most visible on the PDP, it is still being digested by Sparky. It is being digested uh in places and makes it valuable, especially the key features. I think you see on PDPs now. There's that kind of uh almost AI summary or like the key features getting pulled up near the buy box instead of being buried underneath the PDP. So key features, especially is becoming a really valuable piece of real estate. Um, the description, it's there. Uh it's being digested, probably not by shoppers a whole lot. So that's your opportunity to really appeal, I think, to the AEO and the GEO elements and be a little bit more wordy and be a little bit more descriptive of products. Bring those seasonal, those moments, those questions that need to be answered into your description uh and use that real estate uh um well. And I think uh maybe the insider tip is shelf description, something that isn't even necessarily visible in the PDP. It's kind of this back end field that used to be a part of the PDP. It still lives in the Walmart ecosystem. For the longest time, it kind of got forgotten. What I've noticed recently is as you're submitting content into supplier one, a lot of times it requires that field to meet requirements or meet standards now. Well, what are those standards and then how can I meet those? That's another great field for you to put some of that AOGO content, uh kind of maybe even behind the scenes that shoppers will never see, but is readable by agents and algorithms and tools. Yeah, yeah. So there's the insider tip of the day. Use that shelf description that'll never be seen by the human eyes, but behind the scenes.

SPEAKER_00 31:38

So if, you know, as we we're getting, you know, towards the end of this and wanted to bring bring some takeaways. How would you advise someone about the fundamentals in today's environment on Walmart.com? Like, what should they be looking at? What should they be doing in order to be, you know, to work through this transition period of of search to to AI and uh gentic shopping?

SPEAKER_01 32:05

I'd say don't stray too far from the fundamentals. Like the they're there for a reason. Walmart invested a lot of time and energy into building these content quality scores that feel really frustrating at times, but they build the right base and foundation that you can then trial and and error off of. Yeah. Um it's you don't need to start from a like, I don't know, you don't need to start in the wildest corners of this. Like build the right foundation, like start simple, um, and then start kind of trial and erring. Uh, and whenever you're using Sparky, like there, there's no magic code to this. Every brand is gonna approach it differently, every shopper approaches it differently. So there isn't a one size fits all to make sure that you're winning with Sparky. You have no idea what shoppers are necessarily gonna put into that bar. Um, it's not nearly as predictive as searches, where someone's gonna have a kind of a priority term that gets searched the most in the category. Everyone's gonna shop unique to their lives and the way that they think about things. Yeah. So there's there's less of that like specific optimization. It's doing the fundamentals right, it's getting your attributes filled out, it's getting your titles, your description, your key features, your uh your ratings and reviews set. Like it's the fundamentals that we've always talked about. Like, I don't think we're anybody's here trying to like totally reinvent the wheel, which is great that we don't have to change like the but I hear a lot of companies out there saying we can help you with your AEO, GEO, yeah, you know, um optimization and help you rank and and show up and all of that stuff.

SPEAKER_00 33:30

So, you know, I I feel like some of that I think is um a combination of things, which is you know, going back to the traditional shirt search plus you know, the combination of some of the fundamentals that it takes in order to to you know last through this AI transition for sure.

SPEAKER_01 33:46

Sure. And again, I think that probably every shopping agent and every uh retailer and and AI company is kind of approaching it differently. So what you need to do for Google Gemini is probably somewhat different than Anthropics Claude and Walmart Sparky and Amazon's uh Alexa shopping, I guess is is now the new phrase for Amazon. Uh like they're they're probably not a one size fits all answer for those, but if you stick to the fundamentals, that gives you the best jumping off spot for each and every one of them. Right.

SPEAKER_00 34:18

So there were there were actually some some recommendations from the workshop that I wanted to to bring uh bring forward here that were in your presentation. I think these are very practical things that I think brands could do now. Um, you know, first was compared logged in versus logged out results in your product versus a competitor. So logged into Walmart.com, logged out, use Sparky, see how you show up versus a competitor. And these are just you know to kind of see how it works, right? Yeah.

SPEAKER_01 34:51

Um Sparky would be tough because I think that's that's specific to when you're logged in, you'll get access to Sparky. So when you log out, you may not even have access to Sparky, but understanding how personalization plays a part in the search grid. Yeah, I think is really important. Understanding how absolutely the algorithm digests your preferences to service items. Because again, if we're using if we're saying search is the back end finding product method of of Sparky, that personalization goes into the search algorithm that services products. So it indirectly decides what products show up in Sparky.

SPEAKER_00 35:24

Um, you also test broad missions, comparisons, compatibility, price, and branded and category questions. So trying trying those different types of tests in order again to see how your products show up or don't show up.

SPEAKER_01 35:41

Right. Yeah. And that's that this kind of new age of like share of voice, right? There isn't a as clean methodology of creating share of voice, like that. This is the methodology for creating share of model or share of whatever, whatever the right terminology is for this AI shopping world. It is just going out and searching all these kind of different themes and thematic areas and then consistently doing that and kind of collecting the results over time.

SPEAKER_00 36:05

Yeah, you're setting a benchmark, right? Exactly. So, and then document the prompt, the result type, product shown, and sources cited, and the purchase path. So trying to really just document your findings so that you compare it to the next time it happens. And then obviously, you know, if you made changes, what did you change in order to see a different result? Right.

SPEAKER_01 36:28

Yeah, absolutely. And I I think another kind of step in that is whenever you're getting those five item results back in in Sparky, you're gonna have a little button at the bottom of those that's like view more items or or see more. I can't remember what the exact terminology is there. But that will take you to the query that was searched and you can understand what that keyword was, what products returned there. And so you can start to understand, okay, how do I understand this keyword result versus what it returned maybe last time, or how did I search Sparky, or how did I query Sparky differently that led to a slightly different keyword? And you start to understand like what are maybe the primary keywords that are being searched. Yep. And then how does it start to like layer in those kind of supporting pieces? Birthday party, six-year-old, gift, uh, boy versus girl. Like, how does that structure the query, or I guess the keyword that is then searched based on your query and Sparky?

SPEAKER_00 37:19

And then you can use AI to audit those contradictions, but validate before creating more content, right? So you can use AI to actually do some comparison and maybe identify some opportunity areas, you know, depending on the model you have and you know what you're doing. So you know, four, you know, four things that I think brands can do and track over time to see how things evolve, and then they can test different, you know, different changes in order to see what the impact is. Now, of course, things are gonna change on the back end as well. Yeah. So there could be some impact there. So absolutely. Um, all right. Well, appreciate you uh, you know, breaking down, you know, Sparky and Sparky and you know what it takes right now as a you know, this this day on what four o'clock on Thursday, June, whatever. Yeah. So yeah, as of as of this moment on what you know your observations have been and your deep dive, you know, with Sparky and how it relates to the the Walmart ecosystem. Yeah, you know, as we we established early on, you know, this is an ongoing process. We're still relatively early in what this will ultimately become. Yeah. It will take curiosity, testing, experiment, experimentation, patience, you know, a lot of patience, a lot in time, obviously, yeah, you know, to to get it to get it right. So um, you know, so that's a wrap on uh this edition of Retail Media Vibes. I want to thank Blake for coming back second second time on on the show and help us kind of get through a lot of this, you know, thinking of of what autentic shopping looks like, especially with Walmart.com and sp and Sparky. So before we get out of here, is there anything you know you want to share? So we did this, we did this workshop, yeah, right, for a group of a group of people through doing business in Bettonville. Um, if you're a brand or retailer or anybody that you know likes some, you know, like a workshop of your own for for AI or agentic shopping, yeah, uh, you know, Blake and I would be happy to uh do something for you. So you can reach out to me um at retailmedia vibes at gmail.com. So thanks everyone for watching and listening. Uh, if you want to catch up on all of the episodes of Retail Media Vibes, you can go to retailmedia vibes.com. And thank you guys again so much for listening. And as always, I promise to do better next time. B V outfit.