Retailgentic, presented by the ReFiBuy Media Network and hosted by Scot Wingo, is where Agentic Commerce meets retail innovation. We help retailers and brands prepare for the future of agent-driven shopping with actionable news, expert insights, dxis, forward-looking predictions, and practical guidance on Agentic Commerce Optimization (ACO).
Welcome to the Retailgentic podcast where we explore the intersection of retail, ecommerce, and AI agents, which we call Retailgentic. Our mission is simple. We are shining a bright light on the new Agents AI shopping megatrend, which will disrupt the trillion dollar digital retail market. The stakes are high. Google, OpenAI, Microsoft, Perplexity, Amazon, Apple, Meta, and Anthropic are all fighting to be the consumer AI shopping agent winner.
Speaker 1:There are gonna be some tough decisions for online brands and retailers to make. What happens to checkout? How should payments work? Could you continue to invest in retail media networks? What does the world look like when over half of your traffic is not from humans but from AI agents?
Speaker 1:These existential questions are the tip of the iceberg. On this podcast, we're talking to industry experts, analysts, and influencers to get answers to these questions, identify trends, and best practices. This is Retailgentic, your podcast partner in navigating our AI shopping agent future. And now here's your host, Scot Wingo.
Speaker 2:Welcome to this episode of Retailgentic. Today's podcast is very different than anything we have ever done before. And, also, today, you're gonna hear a podcast that is literally the first of its kind. I was invited by Ben Miller, VP of content and strategy at Hive, who is the group that owns Shoptalk, to participate in the first ever Shoptalk debate format. This is they're trying to have some meaningful conversations.
Speaker 2:There's a lot of arguments around various things happening in retail, a lot of disagreements, and they thought it'd be fun to surface some of those. The topic for this debate was Agented Commerce. Will AI agents transform retail? I guess where I'm gonna fall on this one. But on the pro Agented side, we we had, Ecta Chopin.
Speaker 2:She is amazing. She's the chief digital officer at e. F. Beauty, which has has a bunch of brands, and she's very forward thinking and, writes a substack that I, I'll link to on the... My substack of that called the AI Digital Chef that I think you should definitely check out.
Speaker 2:And then myself, so Zetta and I on the pro Agents side. On the anti Agents side we had Sarah Marzano. She is the principal analyst for retail, e commerce, and retail media at eMarketer. We also had Andrew Lipson. He is an independent analyst and consultant for media, ads, and commerce.
Speaker 2:I thought this was... I've had debates with Sarah and Andrew on LinkedIn. You may have seen some of those. And a lot of it comes down to definitional stuff, so we... We'll cover that and cover a couple other topics.
Speaker 2:I thought it was a fun format. I got a lot of positive feedback from the audience. Obviously, my my readers and audience here are very progenetic. So most of it, was believing that that we kinda won that side of the debate. I don't I don't really think of this as a win or loss thing.
Speaker 2:I think it's interesting to hear the naysayers point of view and kinda like understand because in your companies, you're gonna come up with people that don't get this and don't understand how it's transforming things. And my job is to try to arm you with as much, ammunition as you can to make the counterargument. So, before the show, I asked the Shoptalk organizers, Hive and all the participants it would be okay to turn the conversation into a podcast. And, thankfully got approval from everyone. So on on bank, thank Ben,, Sarah, Andrew for, agreeing to participate in this.
Speaker 2:And we have now turned it into this podcast. So we also believe in this new AI era, speed really matters. So this debate happened Tuesday, March 24 at noon Pacific three eastern, and we're getting it in your hands, in your ears two days later so that you can hear this, basically real time. So hope you enjoyed this debate. First ever of its kind and the first only audio, podcast we have out for our video folks.
Speaker 2:We've tried to put a couple things on there to make it interesting, but unfortunately, this one is audio only. Thanks again to Shoptalk and the folks at Hive and the participants for letting us publish this. I hope you enjoy the debate.
Speaker 3:Well, good afternoon, everybody. A very warm welcome back to a brand new session for Shoptalk. You know, we're at an inflection point where the next steps on so many things in our industry are not clear. You know, it's possible to talk to a really smart person, experienced and respected, and hear one side of an argument, and then to talk to somebody else who you really know, trust and respect, and hear something completely different. And we wanted to reflect that ambiguity and that uncertainty in the programming at the Shoptalk.
Speaker 3:So very pleased to know me. I'm Ben Miller. I'm VP of Original Content here at Shoptalk, and I'll be part of the team that's organized some debates for us. We've got three different debates coming up through Shoptalk over the next two days, starting right now on a fantastic debate of whether AI agents will or will not transform Retail. And we've got two great teams to debate this with us today.
Speaker 3:So to start with, introducing the team who's gonna be arguing the affirmative. We have Scot Wingo and Eckert Chopper. Scot is CEO of Agents Startup. Think that's a...
Speaker 4:Pro Agents Startup, let's go.
Speaker 3:I think that's kind of premature celebration, Mr Wingo. So my very last words were, Please behave, which has Scot gone that is CEO of Agents startup ReFiBUY. He's former CEO of ChannelAdvisor. He is co host of the ever popular and insightful podcast, The Jason and Scot Show, and he's now also has a new podcast and Substack tracking the development of Agentsic Commerce called Retailgentic. Hector is Chief Digital Officer and AI Officer at e.
Speaker 3:F. Beauty, and also supports on a number of boards and organizations with her AI insights and expertise, including Ad Age's AI Advisory Council. For example. Arguing the case against, we have Andrew Lipsman and Sarah Mazzano. Warm welcome, please.
Speaker 3:Andrew is a well known industry commentator and has led his independent Consultancy, Media Ads and Commerce, since 2023. And Sarah is eMarketer's Retail and Commerce media expert, and just last month promoted to the role of Vice President and Principal Analyst. So congratulations on that, Sarah. I mean, look, the teams for this almost picked themselves. Scot and Andrew especially have been arguing this topic on LinkedIn for what feels like months now, so I'm delighted that they've agreed without hesitation to join us and settle the debate once and for all in front of you all at Shoptalk today.
Speaker 3:Look. We're gonna be having a relatively structured debate. We're gonna be working through four key questions with space for opening statements, rebuttals, discussion. The idea is to share as much data, insight, opinion and knowledge with you as possible to help all of you come to your conclusions and take away your own views at the end of the day. So look, we're going to start off by ensuring we're on the same page.
Speaker 3:What what do we mean by Agents, by AI, Agents, Transforming Retail? So how do we define it? Andrew, please start us off.
Speaker 5:Well, first, I just wanna say, it's great to be here. None of us on stage is the arbiter of this definition, but there's been a lot of debate over what constitutes Agents Commerce. And there's a very broad definition, which basically says anything that AI touches on the way to commerce is Agentsic commerce. AI existed before the word Agentsic did. And so I think there was already AI assisted shopping and other forms of AI assistance.
Speaker 5:Agentsic to me is defined by autonomous decision making of the agent. So we really need to define and narrow that scope. And so I don't think anybody disagrees, certainly Sarah and I don't disagree, that AI is absolutely changing the way we search. It's changing the path to purchase. But this whole idea of agents just issuing a prompt and setting and forgetting it and letting them do the decisioning and then ultimately leading to purchase with minimal or no human intervention.
Speaker 5:And I'll allow for minimal human intervention. It doesn't have to be end to end. But I just don't think that's ever gonna happen. So to me, that is what I have called a collective hallucination.
Speaker 3:Okay. So I love that we've already got into your pinning your colors to the mast when we're trying to do the definition. So thank you for that, Andrew. Scot, how do you define Agents e commerce?
Speaker 4:Yeah. So I've been in LinkedIn fights with both Sarah and Andrew on this. And I couldn't understand where they were getting this very unique definition they have. Believe it goes back to Stripe in 2024, had a definition that was kind of like that. And I asked the CRO at Stripe, is that how you guys think about Agents and Commerce?
Speaker 4:And she said, where'd you hear that? And I said, Andrew and Sarah. And she said, We did say that in 2024, we no longer say that, we have a much broader view of it. And they actually created Agents and Commerce protocol with OpenAI. So, my definition of Agents and Commerce is the researching, find, and potentially buying of a product off of your website using one of the answer engines that have grown up.
Speaker 4:So we all know Chatbtq, now Gemini, there's five of them today. Rumor is anthropic. I see that they're here in some places, so they're coming out with one soon. So, that's what I think about with this. But as Andrew said, definitions don't matter because what Andrew and I think isn't what's important, it's what consumers are doing.
Speaker 4:And consumer behavior is changing more in the last two years than it has in the last twenty. And there's a McKinsey report that just came out last week conveniently, fresh data from February 26, so this is not old data. And they basically surveyed 4,000 consumers, which is largely n equals over 2,000 is viewed as a good sample, and they found 68% of consumers are engaging in some form of shopping on one of the answer engines. I had Ashish from Gemini Google on the podcast. He said the reason they're doing it is because the consumer data is just irrefutable.
Speaker 4:Consumers are entering prompts and they want the engine to help them. The same is true for OpenAI. Today on the stage over there, they announced some new discovery features. They're not getting out of Agents Commerce, so Andrew is dancing on the grave a little bit, but that's okay. And they're doubling down on discovery and they announced a new comparison function, a new feature with ACP to do better discovery.
Speaker 4:They're really drilling in on product data. And I think the consumers are pulling it. Last point, the people in this room, there's another survey from Logic Broker, they're over there, and they surveyed 600 chief digital officers like an ECTA, and they basically found 61% of them are, 48% of them are investing over a million dollars this year on what I call Agents e commerce. Off-site discovery, finding of products, and even checkout.
Speaker 3:When we were talking before we came on to the stage, you were talking a lot about the change management you're having When to do within you were talking to the business about Agents Commerce, how are you helping them understand what you believe, what you understand by Agents Commerce? How are you defining it with your colleagues?
Speaker 6:Yeah. I mean, I think it all starts with education. We did bring in some external thought starters. So we brought in sort of one of the vendors who's actually really invested in this area to come in and really educate our marketing leadership team. Because when when you think about this, right now we're talking about agents, but it's about brand visibility in those, you know, search engines.
Speaker 6:And if you're not there as a brand, you're gonna lose the game. You're irrelevant. So it starts with that. And I think that is a mindset shift that needs to happen in the organization more than a, you know, are we gonna be able to check out or not? It's it's a broader discussion about how this world is changing and the roles inside the organizations are changing as well.
Speaker 3:Really interesting points to come back to. I'd love to bring you in. When you are conducting research at eMarketer about Agents, how have you come about finding a definition that then when you talk to the industry or you talk to consumers when you're surveying them, that people understand what you mean?
Speaker 7:Yeah. I mean, so I would refute the idea that definitions don't matter. I think they do because when we're urging brands and retailers to make big decisions and make massive investments preparing for a future, we need to be really clear about what they're preparing for. Now the good news about the goalpost moving so often when it comes to defining Agents Commerce is that if all we're talking about is a new powerful discovery surface, then brands and retailers have been adapting to that kind of consumer behavior for a long time. And we have plenty of historical precedent for how to adapt to where consumers are spending time and how we show up on those surfaces.
Speaker 7:That's materially different than saying that transactions are gonna move at scale or autonomous decision making is gonna change who your customer is. And that's real discourse that's been happening in our ecosystem and in our industry, which is why it's really important to get very clear about what we're talking about here. Yeah.
Speaker 5:Can I can I jump in on a data point that Scot shared earlier? So I think that was the McKinsey report. Right? And the numbers in the report showed 23% of people engaged in checkout. 23% of the 19% of people who did shopping, of the 68% of people who had engaged with AI at all.
Speaker 5:If you do the multiplication on that, you get down to 2.9% of people who have actually said that they have done checkout. Now, the real number of how much checkout is happening is probably way lower than that, because this just means you've done it. You've ticked the box once and not that it's a big part of it. So I just want to push back on the numbers because I think a lot of times these numbers, we can cherry pick data that make something sound big, but when you actually unpack it and dig into it, it's actually a very fractional behavior. And if that's not evidence enough, we saw that Walmart walked away from instant checkout, and they stated that it was because the activity was not happening.
Speaker 5:Checkout was not happening in any consequential form or fashion.
Speaker 3:Which I think really nicely moves us on to the second area that we want to talk about, which is customer adoption. But I think what's really clear so far is that for you guys, for it to be pure Agentsic, it has to be a level of automation. Actually, for you guys, what we're talking about is using anything in the Agentsic Commerce stack to make it. So what I think we're going to do for the next half hour is make sure we talk to both those points because I think we recognize we've got two different elements of the spectrum, and we're going to address them both as we go through. So let's start with customer adoption.
Speaker 3:Scot, I'm going to turn to you, especially now refocused by your hands on in helping organizations build their tech stacks for Agents Commerce. What are you seeing so far around consumer adoption of the various elements of Agents Commerce?
Speaker 4:Yeah. I put out a framework that kind of says, I do think we get to some form of autonomy. I don't think it ever gets to kind of like this rough definition of a 100%, because I don't think people will ever shop that way. There's gonna be certain things you shop that way, but but I have this five levels of automation. And so, I kind of look at the answer engines and where they are.
Speaker 4:They all started at what I would call research and then they've added find. And Buy is brand new. Remember, Buy just launched like three months ago. Gemini just has like three people live. So, yes, it is small today, but they're investing a lot.
Speaker 4:And I think this holiday you're going to see some, you know, that 3% number will start trending towards 10% and then so on and so forth. But I think we'll have this matrix of how people shop. There's going be certain things you want the engine to help you research or you may want to do that yourself. Like my daughters, when they're searching for an outfit, they want to do that and then they would love to have an assistant take over the rest. Okay, I want to buy this, now go find me the best price and the fastest way to get it.
Speaker 4:Or you may want it to do some research, present you with options, this is how I got a coffee machine, then I picked one, so I got involved in the middle and then I had it finished towards the end. So, I think everyone's going to kind of pick their own way they use these tools and it's not going to be this, This is what I don't like about this all or nothing definition is the logic goes, our definition is it has to be totally autonomous. Therefore, don't see that happening. Therefore retailers shouldn't do anything. And I think that last step is getting the causality wrong in this.
Speaker 4:I think it will come, and you have to get ready for the front part. What we're seeing is a tremendous investment on the discovery side, making sure you're getting your product data in a format that the engines can use it. And this is below... You have AEO and GEO, which is way up here. That's what I call sentiment analysis.
Speaker 4:As you go down, everyone sees the little product cards and then the the offer card. Most retailers and brands are not showing up on those offer cards because these new engines are not great at mapping things yet. This is exactly what happened at Amazon in 2006 when they launched their marketplace. So that's what we spend the bulk of our time on right now, is what I call... It's a super boring word, but it's canonicalization, is giving the engine the hints it needs and the product data it needs to make that mapping so the consumer sees the product in the right context.
Speaker 4:Then you have to, once you get the mapping, we have this new world that's kind of, we've been in keyword jail for like twenty years where we had four keywords and five attributes, and that was... We were good because we outdid... We had way more data for the product than the shopper intent. Now the shopper intent is like literally megabytes. We know so much about the shopper, but now we have five attributes.
Speaker 4:We've got to beef up, you know, every... We have to add a lot more attributes, but we also have to add a ton of context. And the engines through UCP and ACP have given you a fast lane highway to inject that right into the engine, so now the product context can map the shopper context. Yeah. When you do that, the discovery is gonna get magical because it will know so much more about the buyer and the shopper and make those matches.
Speaker 3:Okay. So that's mapping... That's getting the the the text. That's getting the architecture ready. Angie, what are you seeing or not seeing in terms of customers' appetite to use it and adopt it?
Speaker 5:Yeah. I mean, first off, I I don't know the first thing about the architecture of this. Right? I focus on the the consumer here. And the consumer...
Speaker 5:What's been stated is that consumers will basically allow... It's gonna be bots talking to bots and doing all the decision and making purchases for you and making purchases in anticipatory fashion based on all the stuff that it knows about you. And so that's the specific thing that I push back against. The consumer behavior to me, I liken it to a collapsible funnel, which is that AI has a lot of innovation across the path to purchase. There's more horsepower in the searching.
Speaker 5:It can do more different things more quickly. So it's a collapsible funnel in that it can shorten some of those rungs of the funnel. So the funnel can get shorter. And that's an innovation. But for the consumer to purchase, they still need to pass through every phase of the funnel.
Speaker 5:And that's the part that I think is lost. The consumer gets pulled out of the equation when we're just talking about a bot connecting to a bot through a protocol. I don't know. To me, the logical conclusion of that is that we're basically a bunch humans living in WALL E. And I don't know about you, that's not a future I want to live in.
Speaker 5:I think people shop for a lot of reasons. We are human at heart. And so that's, to me, the behavior that ultimately will be persistent.
Speaker 4:You agree with everything but the definition. You agree that it's gonna shorten the funnel. It'll be new discovery. You just don't... You don't like the bot to bot definition.
Speaker 5:I think there's certain elements, and there's other elements that that I think probably won't change things all too much. Because if you break it down into the component parts, these are all... Comparison shopping. This this has existed a long time. It's not hard to do.
Speaker 5:It only works for products that are consistent SKUs across a bunch of different retailers. And guess what? We're doing this in 2005, and we moved away from it. We moved to Google, and then from Google to searching directly on retailer sites and Amazon. So this behavior is something that consumers largely I don't want say have fully rejected, but it's not that important a part of their process.
Speaker 3:Sarah, help us understand what you're seeing in the research at the moment because I think there is a logic train that is if you follow through the process, then the bot to bot, the automated stuff, the natural end game that a number of repeat purchases will move to. But clearly, it's an argument that Andrew is very echoing. He's pushed up again... Back against. What are you what are you seeing in the eMarketer research that will help us understand where we think this might end up?
Speaker 7:Yeah. I think all of the discourse around this inevitability of a future where consumers are looking to outsource any meaningful share of their, purchases relies on this sort of fundamental misunderstanding of the consumer and and why we shop. Like, people actually really enjoy shopping. It's not a chore, to be outsourced. And we have some historical precedent that helps us prove that.
Speaker 7:Right? Ordering groceries online has never been easier, but more than 90% of grocery purchases still happen in physical stores, which is arguably harder to do and includes more friction. Subscribe and save has been broadly available for years. There have been entire d to c brands that were, invented with that sort of purchase behavior in mind. And what we've seen is that Amazon, for example, has lost share across...
Speaker 7:I think it was nine out of 10 of their subscribe and save categories. It's simply even for those product categories that we pinpoint as being the most likely, to wanna offload or outsource. We're not seeing any consumer appetite for doing that at scale. So I think it's just important to keep in mind what consumers have shown us in terms of, like, what drives them to shop and what they're likely to have an appetite for going forward as we plan for this.
Speaker 5:Quick quick little addition to to that point also. Why they're rejecting it is is is because... I'm sorry. Forget that.
Speaker 3:I forgot the point. Perfect. I love that. I mean, you... You've got, in beauty, often more high consideration purchases.
Speaker 3:There's an emotional attachment. How how are you seeing customers respond to the early things that you're doing at the moment?
Speaker 6:Yeah. For us, as I said, what do I believe in? I do believe in... Like, there are ecosystems that are building around Agentsic. Like, when...
Speaker 6:With Open Claw coming out now, you actually can have an agent that does autonomously think... Things for you, and then NVIDIA making an announcement on, you know, their version of Open Claw, which is gonna be enterprise ready. I I do think that we are at the beginning of what's gonna be built on this whole world around Agentsic. When I think about our brand, we're focusing on showing up there. So in a...
Speaker 6:There... For example, average search is five words. But for for me as a consumer of skincare, now I can go into ChatGPT and and other places and just say, hey, I want a skincare that's for my... This particular problem, and it's... I I wanna...
Speaker 6:You know, I live in a warm weather. Now that's more things that I have to get my product data, my content ready so I am actually in their consideration set. So I do think visibility is absolutely important. Your product data, getting that. I truly believe the architecture of the product data will be sort of the mastermind behind how this sort of works.
Speaker 6:So that in itself is really important. I also think that we have to start preparing for the world of Agentsic. So when I think about CPM, CPM is actually gonna change into, you know, cost per agent recommendation. Because if you do start having Agentsic shopping, you're you're not gonna be thinking about impressions and so forth. You do have to come up and start measuring it in different ways as well.
Speaker 6:So we're preparing for for that world.
Speaker 3:Okay. So just to wrap up this section, 2030, will we see 10% of e com be completely bought to bought, be automated by 2030?
Speaker 6:By 2030? Yeah. I think so.
Speaker 5:Not chance. Not chance. Than 1%.
Speaker 4:Mackenzie Mackenzie says so, Arc Invest, Morgan Stanley, Deloitte, and several others. So
Speaker 6:And latest announcement with Sephora just integrated their loyalty program with ChatGPT. That was their late... Latest announcement. I do think this ecosystem is bringing in some of the, you know, bigger retailers into the ecosystem, and I do think that's gonna start.
Speaker 5:Consultants create demand by by these big forecasts. Right? That's part of the playbook. So that's not to me evidence. Like, they don't know it as much as anybody else is.
Speaker 5:So I think we should take that with a grain of salt. And announcements? I'm sorry. Like everybody comes up with announcement in advance of these tentpole shows. How many announcements that you've heard over the years actually materialize into something?
Speaker 5:The rubber hits the road when you have to get commercial teams together, you have to build product, build the pipes. Like all these things are very complex things, and all of them have to be true in order for something to materialize, the behavior to materialize and then the economics to materialize behind that. You know, I'm gonna reserve judgment until I see it happen. Okay.
Speaker 6:Yeah. Can I add one point there? I do think it's true probably for big enterprises that create those barriers. I do think we operate at e. F, and maybe that's our advantage, as a...
Speaker 6:You know, even though we're like 1,500,000,000, we we definitely operate as a big startup. And I think that mentality is what really is gonna push the needle. If you're still stuck in your enterprise sort of, you know, silos, I don't think you can move this forward. I
Speaker 3:wanna come back to what retailers and brands should be doing now. But before we do that, I wanna kind of take a side route into Retail Media. There's been a lot of conversations recently about the relationship between Agents and Retail Media. Ultimately, do agents create an existential risk to the Retail Media model because the traffic's going? Sarah, kick kick us off.
Speaker 3:What's your view here?
Speaker 7:Yeah. I'm glad to get to weigh in here. Look, I imagine at least half of you spend as much time on LinkedIn as I do. And if if you do, you've probably come across a really, scary line of thinking, that goes along the lines of by twenty thirty, retail website traffic will go to zero, and the website as we know it will be a vestige relic that catered to a time when human beings shopped. And that's terrifying for anyone in this industry.
Speaker 7:Right? It... It's it's enough to send chills down your spine unless you spend any amount of time thinking about it critically. So let's do that. We have a fair amount of evidence as well as historical precedent that shows us, that traffic to retailer site is not in danger and that retail media's value is not going to be undermined.
Speaker 7:Let's start with, whether or not traffic is going to decline significantly. Just want everyone to keep in mind that the only way that's going to happen is if transactions move off of retailer websites at scale. Okay? Now out of the 144,000,000,000 that eMarketer is forecasting for AI platform driven ecommerce, we have about 28,000,000,000 of that slated to happen within... Natively within AI driven apps.
Speaker 7:That's about 2% of all ecommerce sales. So traffic isn't going anywhere, again, unless transactions move at scale. And this is where the really important historical precedent comes in because we have a long history of tech platforms who do really well at gaining consumer attention or playing in the discovery phase, looking to extend their influence into transactions. Some of them you might have heard of. Google, Meta have all tried to move into owning the transaction layer and largely failed, which really shows us that owning attention does not equate to the right to own the transaction.
Speaker 7:So let's put the fears about, traffic disappearing overnight to the side, and let's think through how, AI, discovery and referrals are reshaping how consumers show up on retailers' websites. We have data there too. We know from SimilarWeb that when we look at referral traffic from ChatGPT compared to referral traffic from Google, that these consumers are entering retailers' websites more engaged, with higher intent, and that they're more productive. Right? They spend more time on retailers' websites.
Speaker 7:They look at more pages, and they convert higher. Right? Two points higher based on the last analysis that I saw. All of this is major upside for retailers and retail media. And I will stop talking in a second, but I do wanna weigh in on another popular narrative that's gaining steam, which is the idea that as referral traffic from, AI platforms evolves, it will increasingly deposit users lower in the funnel on retailer websites.
Speaker 7:And the argument is that, therefore, the value of retail media will go away because those consumers will be less exposed. To believe that, again, you have to put aside everything you know about how consumers love to shop, and you have to put aside the ten years of history we have around referral platforms increasingly depositing users lower funnel on retailers' websites. We know that the consumers who are hell bent on transacting on one item are gonna do so, and that the majority of consumers who are open to discovery are gonna continue browsing and basket building, which good news for retailers, they're really good at it. So again, buying into this narrative, requires you to completely ignore the way we know consumers like to shop, and it also requires this, like, willful ignorance of all the well documented issues that the AI platforms have had as they've tried to get single item checkout off the ground. So retail media does not face an existential threat, neither do retailers' websites.
Speaker 7:In fact, there's a lot of upside to be considered.
Speaker 6:K. So just a question there. Then do you put all the consumers... Because there's multiple consumers. There's Gen z, Gen alpha.
Speaker 6:When you say consumer, which consumer are you talking about?
Speaker 7:That's right. I'm talking broadly about consumers and how we've seen they like to shop. And there is no misconception on my end that retailers and brands need to segment and plan based on the type of consumer they're catering to.
Speaker 3:Right?
Speaker 7:So strategies need to evolve based on the shopping behavior that they're seeing. But the point contends that, that the point remains that basket building is a huge part of shopping on ecommerce websites. Yeah. And there are some D2C brands who are probably very used to single item checkouts, but most major multi brand retailers are very accustomed to basket building.
Speaker 3:Yeah. Okay. Tell me, how do you think about it? Obviously, e. F, big investor in advertising.
Speaker 3:You think a lot about demand creation, big investor in Retail Media. What are you thinking when you look forward about the amount that you might be investing, the proportions, how important Retail Media will become as Agentsia Commerce increases?
Speaker 6:Yeah. Relatively, to be honest, we don't invest a lot in Retail Media. The only ecosystem that we invest in is really Amazon to a certain extent or our direct channels. We do do testing and learning. And I think that goes back to protect...
Speaker 6:Like our price point. E. F. Is, you know, value brand. And for us, it does not make sense because we...
Speaker 6:Our prices are so friendly anyways. So for us, it's really showing up on those retail websites and so forth the best enriched data as possible to make sure that the consumer has everything in order for them to shop our product. So education becomes super important. Content becomes super important, less so in media.
Speaker 7:But coming back to the original question, do you believe that traffic to retailers' website is gonna decline in a significant way because of AI platform behavior?
Speaker 6:I I think as retailers figure out how they're integrating into these answer engines, I do think it will decline. I don't know to what extent to... Like, it's gonna be completely gone. No. I think it's just gonna be diverted, and I think it'll become a meaningful channel.
Speaker 6:I don't think it's gonna go away completely.
Speaker 7:So it sounds like we agree that some of the upper funnel pieces are gonna get more fragmented as discovery shifts and change, but the traffic to retailers websites is going to remain healthy, and with it, retail media, which is where consumers are exposed to it.
Speaker 6:I I believe that.
Speaker 3:Yeah. Cool. How do you
Speaker 4:say it? I I think it already has declined. So as people get answers at the answer engines, they're getting less and less informational searches. And I've been talking to CPG brands, chief digital officers, and then also a fashion one, and they have a measurable hole in their plan this year because they're losing SEO traffic because the answers are happening off their website. I don't know about...
Speaker 4:You guys know way more about Retail Media Networks than I do, but we are seeing, you know, and then there's the similar web data shows Google referral traffic to retailers is down 10%, but it's hitting and it tends to go by generation. So it's the earlier female generations. So fashion and beauty are feeling it the most right now. And they're feeling a material decline in traffic. A CDO at a CPG said, we've had to get our head around that the clicks are going to die.
Speaker 4:They basically said rip clicks. That's not coming from me. That's like, you know, and they've they've done this trend work, and they basically see the line. You know, I don't know if it goes to zero or not, but there there are a lot of people freaked out by this decline of traffic, and it's creating a material hole in their plans this year.
Speaker 5:But I've got my
Speaker 6:that that does go back to how you're structuring structuring your product or your content. So if you do not pay attention to that structure and the change management around sort of your teams that are currently operating in a different way, not to the answer engines, I do think it's that decline continues unless you're
Speaker 5:of taking advantage want to push back on the substance of the decline. So I've done a lot of analysis of sensor tower data in a bunch of different ways. One, traffic writ large is not declining at all. For large... The largest retailers, 60 to 70% of their traffic is direct.
Speaker 5:Right? So that's non referred really in in the first place. So that's pretty defensible traffic that's not gonna go poof tomorrow. And then the most interesting thing is that if you look at a bunch of retailers plotted, their year over year growth in traffic versus the percent share of their referral that's coming from Gen AI, there's a positive correlation. Now, if what our opponents say is true, then you would expect to see the opposite.
Speaker 5:Right? That the traffic's declining when Gen AI referral is high.
Speaker 4:There's no Gen AI referrals because the answer happened off-site. There is no traffic. Traffic is dead. There will be zero referrals.
Speaker 5:Then then why are they growing? And in some place... In some cases, by considerable amounts.
Speaker 3:So are are we seeing kind of category or sector differences? I mean, Obviously, talking very general, and everybody in the audience will be able to see what's happening to their websites and their Here's
Speaker 5:the interesting. Gen AI referral is highest for retailers in high involvement categories. So think Best Buy, Home Depot, Lowe's. Consumer electronics, DIY, home improvement, things of that nature. And so the funny thing is they're growing.
Speaker 5:They're growing. So people are more likely to engage more deeply in the search process. There's new use cases for discovery and search, but you're still gonna end up at the site, and you're still gonna make your decisions within the context of the site. You're not just gonna outsource it from the decision from the LLM.
Speaker 6:If you... Those retailers that you took names of, one thing in common there is education. So all of them require so much, so much consumer education. So I do think that probably is the reason why, because when you think about discoverability, that's already starting to happen in answer engine. You're going there.
Speaker 6:You're, hey, compare this to this. Can you give me options for this? This is coming because they have robust education information that's coming. And I do think that shrinks the intent. So even though they might purchase from the website, that traffic doesn't go away.
Speaker 6:The actual engagement is happening on the answer engine.
Speaker 5:Why does that traffic come from the retailer's site versus Google? Because there's been an argument that it's getting pulled upstream. And again, the evidence does not support that. There's all of this... You know, the whole path to purchase already involves many, many platforms.
Speaker 5:It's social. It's search. Yeah. It's on-site. It's all these things.
Speaker 5:And so for me, I agree with you. The behavior is changing, but to me, it seems like this large pool of activity that's happening on Google right now is the stuff that's most likely to shift to the use case you're saying.
Speaker 3:Okay. Let's keep moving the conversation on because the time is progressing. Hector, I'd like to come to you first. What advice would you give? What are you doing internally when you're thinking as a brand owner, next six to eighteen months, what should you be doing?
Speaker 3:And then thinking further out the next three to five years from how you are seeing Agents and Commerce evolve, what do you think people in the room should be doing in their organizations now, today?
Speaker 6:Well, number one, it's about sort of really thinking about how it's gonna impact the teams that you have internally and how the work they do is going to change as this shift is happening. So educating your teams around how... Well, how do you think about enriching your product information or your content? It starts with education. Yeah.
Speaker 6:We started doing that almost a year ago. Fast forward, there's work streams around, you know, what are consumers' questions are they asking so we can build that content. We can enrich our product around that. So I think starting with that should be the most important thing. And then doing testing and learning around checkout.
Speaker 6:Do you connect with an API? Do you, you know, leverage those things? I I still think it's in a learning phase, but it's happening. So you do have to start with changing the mindset of your organization to prepare for the scale that will come as these ecosystems are building, and they're building very fast. The pace of change is far greater than what it was before.
Speaker 6:So that's what I would tell people.
Speaker 3:Sarah, is that is that advice you agree with, disagree with? What did you came up to telling its clients?
Speaker 7:I was... I... And I was preparing for this. I'm not surprised that I actually agree with a lot of what Hector said. Right?
Speaker 7:She sits in house at a major brand that's also a retailer. She's going to be pragmatic because she knows that there are consequences to rushing into things. Right? So I'm not gonna spend a ton of time, detailing out all the pieces that I agree with. I'll just say that, you know, retailers and brands should absolutely be testing and learning within these environments.
Speaker 7:They should be thinking through how they show up, but they should crucially not be convoluting the difference between preparation and a massive reallocation of budget. Right? So do not run out, and build an expensive app for one of these new, environments, which is gonna be your next call to action. Retailers, you have to keep your eye on where the greatest upside is, and for now, that's retail media. I mentioned already that we expect about 28,000,000,000 in a native Agents transactions by 2029.
Speaker 7:Compare that with the 50,000,000,000 in incremental spend we expect in commerce media. I'm gonna make you guys do some math, but I'll walk you through it. If we look at the profit pools, let's assign about a a 10% margin for ecommerce, which is generous, and a 50 to 70% margin for retail media, which is conservative. I'm trying to play fair here. We're talking about profit pools that are about 3,000,000,000 for Agents commerce, native transactions, versus somewhere in the realm of 30,000,000,000 in retail media.
Speaker 7:Right? So, most retailers are still really early innings when it comes to retail media. Absolutely keep your focus there. But one thing I love here is that we don't actually have to debate all of this. In in theory, we have real evidence of what happens when we listen to the most extreme case of recommendations.
Speaker 7:Right? We can look at what retailers and merchants were guided to do going into the holiday twenty twenty five season. Following the launch of ChatGPT's native checkout, retailers and merchants were told that they were facing a theological decision, that they had limited time to act, that native checkout was a no brainer. And in no uncertain terms, they were told that the first movers in those spaces were gonna gain market share that was gonna be irreconcilable for the second movers. Right?
Speaker 7:And we have the benefit of hindsight now. Andrew already mentioned that Walmart themselves said that the conversion rates for ChatGPT's native checkout were three x lower than other referral traffic. We see that, native checkout as it existed within ChatGPT has been completely taken off the table. Right? And this was the reality that retailers were being told, like, you have to act now or else you will miss out.
Speaker 7:Everyone in this room is intimately familiar with the retail industry. That's why we're here. I I don't think you need me to explain to you the risk that comes with taking your eye off the ball during the most important holiday shopping season of the year. And and this is important because it's a lesson that's going to keep coming up. Right?
Speaker 7:We know now that advertisers who rushed to reallocate budgets to show up in the early advertising pilots on AI platforms are now frustrated because they pulled budget from search and social, and now they have money tied up in platforms that aren't performing. And we also know that the next call to action for retailers is gonna be to throw all of your resources into building apps in environments that consumers haven't shown any appetite to adopt. So the lesson is test and learn, but do not reallocate budget at scale until you see proof points for doing so.
Speaker 3:Right. Two and a half minutes left. We're gonna have to wrap this. Scot, I'd love you to give us a minute and a half to wrap up your views. You can respond to Sarah's comments at the same time.
Speaker 3:And, I will turn to you to wrap up as well. So Scot, give us your give us your closing argument.
Speaker 4:Yeah. I agree. Retail media network has a lot of margin. I think it hurts the discovery process to to have so many ads in there. As a consumer, I'm not a lover of all these ads getting between me and what I'm trying to find.
Speaker 4:So have a balance there. Don't overdo it. I think Amazon's tipped way over to overdoing it. The Sensor Tower data is largely on Android, and it favors Amazon, which is... And Walmart, which is All Direct.
Speaker 4:I would look... I would not use that data for this purpose. I'd use SimilarWeb. I think it's a better panel. And I think it's pretty irrefutable that the consumer is using these for discovery.
Speaker 4:And I think that it is a new world. You can't compare AI today to 1998 AI. In 2022, the introduction of LLMs via ChatPT has changed everything. You can have... When you have that experience, go experience it yourself and compare it to Google Shopping and I think you'll disagree with Andrew that we've done shopping before.
Speaker 4:It... It's not gonna work this time. It is fundamentally different. It's a multi turn this discussion. It's like having the best sales rep, online now.
Speaker 4:And I think it's actually gonna take share from physical retail. And
Speaker 3:Jake?
Speaker 5:We've already moved the discussion from, full scale Agents to Commerce to getting back to Discovery. And again, we don't disagree about any of that. Discovery is evolving. I think, you know, a final point that I would make is that I don't think any retailer should be making decisions based out of fear. And there is a whole industry and ecosystem that is trying to compel everyone to make these decisions out of reflex because they're fearful.
Speaker 5:Now, part of the argument behind that has been that we've never seen anything like what's happening with ChatGPT. There's this exponential growth. If you don't do it now, you will be left behind. How many people in this room are aware of the fact that ChatGPT over the last six to eight months usage is flat? Weekly average users, flat.
Speaker 5:Total sessions, flat. So this exponential curve, right, we like to think it goes like this. Oftentimes, more of an s curve. This one just ran into a wall and it's flat. Not saying it's not gonna grow from here, but I think we need to stop the hysteria and and, get real about what's actually happening here.
Speaker 3:Brilliant. Look. We're gonna wrap it up there. And like any good debate, we weren't gonna settle it here, but we'd love to hear your views. If you violently agree or disagree with what you've heard, we actually have got one of our camera crews out.
Speaker 3:We'd love to capture some sound bites. Stop with my colleagues. Tell them if you agree. Tell them if you disagree. Come and give us a hard time afterwards or tell us how you completely agree, and it was a breath of fresh air.
Speaker 3:But please, most importantly, thank these amazing colleagues for their conversations. Thank you, everybody.
Speaker 1:Follow, rate, and share to stay up to date with Retailgentic, your guide to the future of AI shopping agents.