Subscribe
Share
Share
Embed
Unlocking Retail Media is the essential podcast for leaders and marketers navigating the rapidly evolving world of retail advertising. We move beyond day-to-day operations to explore the strategic future of the industry, covering major investment trends, the shift to hybrid marketplace models, and the existential disruption posed by Agentic Commerce. Host James Avery brings in top industry veterans and visionary founders to analyze how ground-breaking technology is transforming customer journeys, influencing product catalogs, and forcing retailers to rethink on-site, in-store, and digital media strategies to remain competitive in the modern age.
Scot Wingo: [00:00:00] The nail in the coffin they use is, and Amazon and Walmart owned that. So this is not going anywhere. And it's like you can have that opinion and you can use that definition. I'm gonna keep calling it agent to commerce and if you, if you want to fight me on that, we can. I don't care. What bothers me is giving people the advice to ignore this because it's, it's, that is really bad advice.
Yes. So I, I do not take that advice. If you're in retail media network, you should definitely be concerned about this.
James Avery: Welcome to unlocking retail media. The podcast where we explore the evolving world of retail media, from data strategy to monetization and everything in between. This is the podcast that breaks down how retailers can build smarter data-driven media networks by aligning with what brands truly need from scalable ad solutions and meaningful metrics to cross channel attribution and programmatic strategy.
Welcome to unlocking retail media. Uh, today I'm joined by Scott Wino. Uh, Scott and I have known each other for, for over a decade now. [00:01:00] And over the last, uh, two episodes. Hopefully you've listened to those already. Uh, talking with Mark Grether, talking with, uh, Matt Nichols. Something that kept coming up was AI and agentic commerce, and, and I thought, you know, what would be better than to get the guy who I, who I think knows more about this than anybody else, at least that I know.
Uh, and kind of what's going on in the AI and commerce space. And then really we can dig into how we think it'll affect retail media. Uh, but to kick it off, Scott, you want to give the audience a little intro about yourself.
Scot Wingo: Yeah. Yeah. Thanks James. Thanks for having me on. I'm always excited to talk retail media networks.
Um, uh, it, it's funny, I actually got in a fight recently with a bunch of RMN people, I'll tell you about that, uh, a, a LinkedIn fight, so it wasn't like a real fight by any means. Uh, so I am a serial entrepreneur here in the Research Triangle Park, like you and, uh, I've, I'm on company number five. The one that's most relevant to what we've done is my third company, which was called.
Channel Advisor, uh, I have an engineering background and what we [00:02:00] did at Channel Advisor is help brands and retailers sell on marketplaces. Um, we started that in 2001, if you can believe it. Uh, and we, the first marketplace we worked with was eBay. And, uh, we thought we had, I had this vision that eventually there would be all these places you would want to sell products off of your website.
Um, and there was only one in 2001, I was right, but it took six years for the next one to come along. So it was a very long, long arduous wait. Uh, and then in 2007, the second marketplace showed up and it was a pretty good one. It was worth the wait. It's this, uh, bookstore called Amazon, and they decided they were gonna really lean in on this, uh, and you could look back and the pro prosperity and the, uh, ascendancy of Amazon is largely tied to that marketplace, this third party marketplace.
Um, once they did it. It was, uh, off to the races. Everyone was opening marketplaces. They became very popular in the APAC region. So everyone knows Alibaba and Mol and all those guys and every, uh, country in Asia, [00:03:00] uh, had one. And then South America with Marco de Libre. Then back here in the US every retailer started adding a marketplace.
And then Amazon started innovating on ads with their ad piece. And that was really interesting. Uh, it was kinda like adding fuel to the marketplace fire because it gave both brands and the third party sellers a way to supercharge their sales. Um, and then also for the marketplace operator to make even more money.
And, uh, since then we've seen Walmart Target, and now, you know. Every grocery store I saw Best Buy is there. Ulta has one. I think Sephora, you know, everyone is opening a marketplace because, and then shortly thereafter or in concert with the opener Retail media network. So then just to kind of finish that story, um, we went public in 2013, and then in 2015, after a couple years of being public.
That was fun, a fun run. And I had realized, uh, I don't wanna run a public company, so I kicked myself upstairs. Uh, so I was, uh, on the [00:04:00] board, started another company, and then we ultimately sold Channel Advisor to private equity and they merged it with another company called Commerce Hub, and they rebranded that to Rhythm R-I-T-H-U-M.
So I'm no longer associated with that. Then, uh, did a company in the middle there and then, um, in the fourth quarter of 2024, I was, uh, and we can talk about a little bit more about this if you want. I was curious about AI and what was going on, and, uh, started reading some things and then decided to start doing some experimentation and then I came up.
A lot of things happen in there that we can go into, but I came up with this theory that we, we were gonna build some stuff for Marketplace World 'cause I, we could solve some problems that hadn't been solved before. And then Perplexity came out with Buy On Pro and I was like, wait a minute. I, I think these, these answer, they were lms, then they became answer engines.
I think these answer engines are gonna become marketplaces. They're gonna be a place where people go [00:05:00] to shop. Uh, I called it agentic shopping. Um, I was off by one word. It, it, you know, the name that is like, really solidified is agentic Commerce now, which is fine. I'm, I, I don't care what words we use. Uh, and then that has, unlike my first experience of predicting this stuff, it happened within eight months.
So we now have, you know, chat, GPT is a full bore marketplace. You can check out there and whatnot. So that company's called refi Buy. And that stands for three words, shortened, jammed together, research, find and buy, which is the shopper's journey. And the inspiration for that, uh, is two things. Number one, we feel like this shopper's journey, which is like really complicated today, is just gonna get jammed down in, in a good way for the consumer.
But for everyone else in our world, the merchants, the, the, uh, you know, the operators, the what you would call publishers. In your RMN world there, that's gonna be like a new world that we're gonna have to navigate. Um, and then also that, that's a throwback to, I had a, a fun conversation in the early days of eBay with Pier, uh, one of the founders.
And he talked about, you know, that was their vision was to help people [00:06:00] research, find, and advice. So I always, always thought that was a nice succinct way to, to say it. So it's a little homage to Pier o. Awesome. Yeah, so formally started the company in January and we're super early. Just came out with product and getting first customers going
James Avery: and where, so you were, I mean this was end of 2024, we were seeing chat TBT kind of start to come out there.
We were seeing, you know, I guess it was getting traction. You were starting to see some of these ordering pieces, like how, you know, taking a step back, like how prevalent do you think ordering through? Chatt BT actually purchasing like end-to-end purchase through, through a chatt BT or a perplexity. Like where, where do you see that fall in the, the adoption curve?
Scot Wingo: It's very quick. So number one, what's fun about ai? And you know, I've been at this technology stuff for 30 years and I've been through the mobile revolution, all, all, you know, all mainframe, all the whole thing. And, uh, never seen anything this fast, which is fun. I like to go fast. I know you do too, [00:07:00] so it may be too fast.
It's like the first time I've ever said that, but it's really fun. It causes you to adapt in ways you never thought you would have to. Um, so, so the progression there, so Chat GT came out in late 22 and it, it was pretty bad. It was kind of like, you know, it was mostly hallucinations, but you could kind of see there was something there.
Um, interestingly enough, I was like, interested in these GPTs. Early on I'd read the Google paper just 'cause I'm a nerd. And then, you know, once chat t came out, it really. Didn't have access to the web, if you remember. Yeah. And it wasn't until there was some hacky ways, like they had a GPT that would do it, but it wasn't very good.
And there was a way you could get to the Bing things, which you had to press six buttons. It wasn't good either. It wasn't indexing it, so it was super slow. Then they added, um, it wasn't until, I think it was March. April or May of 24 when they added the web search capability. Right. And perplexity has been ahead of step or two on these.
So if you are on perplexity, you could kind of see it coming. 'cause it just felt like that was the logical next path. And then once they added the web search, then you could do research. [00:08:00] But then it like just fell on its face after that. It was really good at research though. I was just writing an article.
These guys said you can't use it for high considered purchase. And I was just looking for like a $60,000 watch. I don't wanna buy one. But I was just proving it's actually super helpful for that because if I know nothing about that category, it's really good at helping you learn things. Right? That's like what is really good at, and so, uh, so it was, so I would say it really kind of started up in research once we had web search.
Then somewhere around, um, right when Perplexity came out with Buy on Probe, they started moving to what I would call the find segment where they would pop up these product cards. Um, and this is the experience you see today. If right now they're very aggressively showing you product cards, but if you knew what to do and you were working at this pretty hard, you could get it to show you some products.
And I thought it was kinda like what we would call a comparison shopping engine back in the day. And then it would just, then it stopped and you would just jump off to the retailer and maybe it would have some pricing information. Uh, it was pretty bad. It was like really janky, but it, you could see someone there was working on it.
And then it wasn't until Chacha PT five came out, the product cards took a [00:09:00] huge elevation up and they got smarter and better. And there's this important thing in my world called canonization we can talk about if you're a nurse in it, um, you could tell someone there is working on the canonization, which is a, a hard problem.
I'm extremely familiar with. I've got a lot of scar tissue from this one, so I could tell. Y you know, you and I also talked to a lot of startups. You can kind of tell what's going on on the surface a little bit if you read the tea leaves. So I was like, someone there is really caring about this and there's people working on this.
So then in July, the Financial times leaked. They're working on a checkout. Then I was like, aha. And then that just came out, you know, on the nine 30 is when that came out. So, yep. You know, literally less than 30 days from when we're recording this. So that completes the buy. Part of the thing. And then over that time, even since chat GT five in July, the find portion's been refining little, little small things, but we spent enough, now we have agents watching this and we can kind of go back in time and we can now see this neat progression doing this and see what's going on.
So that's a long [00:10:00] answer, but that's the progression of how it went from. And if we go back to when I, you know, so basically July of last year it was okay at research and then now they've closed, this research has gotten really tight and good and find is getting good and buy is still, you know, it's only Etsy right now, but the rate that they're progressing this by next July, I can't even fathom what some of the, I think having it go buy for you is where they'll do a lot of things next for certain types of products.
James Avery: Yeah, yeah. And I think that's a good point. Like a lot of the conversations I've had have been. Who's gonna be most affected by this? Right? Because we, you know, I think about it in the sense of like powering retail media networks, right? The advertising, the dollars all come from people shopping on that retailer site, right?
And I think we just saw, Walmart just announced they were gonna be in this kind of buy direct from Chatt PT. You know, if you're checking a box and checking out on chat GBT, then you're not seeing any ads on Walmart bar. Right? Nope. Now OpenAI can definitely go build an ad [00:11:00] platform, which we all think they will.
Um, but that's very much more a Google ad platform than it is something for each retailer. Uh, but I also think it's gonna be very uneven in how these purchases happen. Like your $60,000 watch, I would definitely use Chatt BT to go research a 60,000 watch. Understand what are the different brands, like what is this good for?
Like what, you know, how should I think about chopping this? I can't imagine a lot of people are gonna buy a 60,000 watch through a buy button on chat tv t.
Scot Wingo: Maybe they would, uh, right. Like I know people that would, but it is extremely rare. Yeah.
James Avery: Yeah. And the same, I at a conference, somebody talked about this with cars where they were like, you know, you're gonna buy your next car through chat, bt, and I was like, I highly doubt that.
Right? Like I am, I am very most likely gonna go research at different, you know, different types of cars and go test drive them. And to make a 60, $70,000 purchase, like, it's not gonna be an impulse buy on, uh, on chatt. But if I'm ordering golf balls and I'm saying I want to get, you know, title Xs, AV Xs, and I want, you know, the best deal on 'em, [00:12:00] then yes, I would quickly click a buy button and get it right versus going to research this more at the store.
Scot Wingo: Yeah. What. I agree with you a hundred percent, but there's always these interesting little cases, so watches of that genre end up being basically collectibles. Right. Um, and uh, I've, I collect a bunch of stuff. I don't know if you do. I don't collect watches. I veer towards the super nerdy stuff. So like, let's say some Star Wars stuff or something and they're not worth $60,000, but, you know, not yet.
They will be if I just hold them long. Now, the, uh. The other day I was looking for something and you know, I was fully expecting to hop off and it wasn't available anywhere. And then Chacha Petit, interestingly, this is the first time I've seen it. This is like. 10 days ago, it said, these are outta stock everywhere.
Do you want me to check on this every day? And I said, sure. So then the next day at like nine o'clock, it sent me a little note and said, still not available. And then it said, do you want me to check every 15 minutes? I was like, well, that's a big jump from once a day to every 15 minutes. And then I tried the 15 minutes thing and like literally every 15 minutes it's like, not available, not available, not available.
I was like, [00:13:00] I was like, can you just tell me when it's available? And it's like, mm, no, it actually, it couldn't, that. That was it. But you can tell someone's like, actually thinking about this. So you know, it has use cases that I think are gonna surprise us even in these things where you would say, I would never do that.
Like the car thing. You know, like let's say you do the research and you're looking for a pretty specific color in a forerunner, add this experience and it's a total pain to go check the websites, to look for inventory around you. And they really don't want to like. Go to the dealer and have them pull one from somewhere and all that.
Right. Just like want to be able to go like see it, see the color and everything. Like I could have it just scan that for me and you know, I'm not gonna buy there. But the assist lift I get from it is going to increase over time. And to your point, that's time I'm not spending on that Toyota dealership or toyota.com or you know, or Carvana or, or you know, or on a retailer.
James Avery: Yeah. Or if you're looking, I mean, there's big markets in the used car. Marketplaces, right? Yeah. Like if you're saying, I want a used Honda Accord [00:14:00] with 50,000 miles that's white or black and is in the Raleigh area, and it could just find that and tell you point you right to it. You're not browsing that site, you're not looking, you're not trying to get tempted at the new Honda Accord versus the used one.
Scot Wingo: Yeah, and the arguments I get in with people are like, they're like, yeah, but that's not agenda commerce. But you have to realize it is so much better than anything that exists. Right? So to do that search, you're gonna have to do Carvana, probably, uh, what's the other one out on Glenwood Avenue Car. It'll come to me and there's like
James Avery: cars.com, there's, there's a lot.
Yeah. All
Scot Wingo: the onlines, and then you really kind of want to check the one that has the big lots, you know? Oh, CarMax. Or CarMax, yeah. Yeah. So there's, there's like 10, you're gonna want to keep an eye and it's just like, there's no site that will search all this and Right. Because they, you know, but then, you know, chacha Pete can, and, you know, so that you, so it's really interesting.
Um, yeah.
James Avery: So, so when you say it's not, you said a lot of people say it's not a agentic commerce. The other thing I can think of, as you could call it agentic commerce, would be. [00:15:00] Where I'm actually tasking an agent to go just do something for me. Right? Like I'm like, I wanna make fajitas tonight. Just, just go have these at my house.
Right? Whether it's the card or K, Kroger or whatever.
Scot Wingo: Usually it's retail media network people. They write these articles that say, um, agent commerce is all hype and not reality. So that'll be the setup. And then as I read it, I'm like, what, what are these? I like use this stuff every day. It's not just for work.
I find it like so much easier than than Google. Um, and even Amazon, you know, the, the Amazon search experience, I think we could all agree is degraded to being basically useless. And it's, uh, we can talk about why and, uh. So then I start reading their arguments and what they get hung up on is they have this logic path.
They're like, my definition of agen to commerce is it has to do everything for you. And they point to some, you know, they have like three references for that definition and one of 'em, Salesforce. But I've talked to people at Salesforce and they, they're like, I don't know where that comes from. So I, I don't know.
But then they have this logic chain that says, therefore there's only these types of [00:16:00] goods. You're only gonna basically get toilet paper with it, therefore everyone else is safe. And I'm like, Hey everyone in retail and brands, you don't have to worry about this unless you sell toilet paper. And I'm like.
I'm, I'm being, I'm exaggerating with toilet paper. Yeah. But they put it into this category of low consideration, low price. They're like, it's basically reordering stuff. And then they say, and then, and then the, the coffin, the, the nail in the coffin they use is, and Amazon and Walmart own that. So this is not going anywhere.
And it's like you can have that opinion and you can use that definition. I'm gonna keep calling it agent to commerce and if you, if you want to fight me on that, we can. I don't care. But I, what bothers me is giving people advice to ignore this because it's, it's. That is really bad advice.
James Avery: Yes.
Scot Wingo: So I, I do not take that advice.
If you're in retail media network, you should definitely be concerned about this. If you're a brand or retailer, you should, and you and I talked to a lot of 'em, it's funny, like that same progression I walked you through, that's their progression of freaked outness. And right now they're at like an 11 out of 10.
So every retailer and brand is very much awake about this and trying to figure out what to do.
James Avery: So [00:17:00] digging into that, let's say I run a successful RMN, right? I have a lot of onsite inventory that I'm selling. You know what some percentage of those online sales are gonna move to agent commerce, either through the stuff we've talked about or all the stuff that's gonna get built in the next year.
What's your suggestion for them? Right. And how do they either protect that onsite inventory? How do they adjust? What's the new world gonna look like for them?
Scot Wingo: Yeah. I would say it's a call to action and what you have to do is. Just kind of honestly look and say, why are people gonna use this? And I can show you it's a better experience.
And that means you have to match that experience. So, you know, what does that mean? Walmart and Amazon are trying to answer it with Rufuss and Sparky and, and I think that's interesting. I would definitely be looking at that. And then how do you integrate Armin in that? Rufuss has ads in it. I don't know about Sparky yet, so I think they're gonna have to take a step back and say.
I would almost [00:18:00] like, you're not gonna like to hear this. I'd like put arm into a side for a second and then figure out how do we, how do we elevate the customer experience? Like how do we help people research, find and buy stuff faster than the current, you know, the, uh, I've been at this in e-commerce.
Since like 98 and the big innovation is filtered navigation and it is just so hopelessly broken that it's totally useless and we've optimized for keywords so much. The search box is broken. So that has to change and that's what Rufuss and Sparky are doing. They're not just a chat bot, they're trying to underneath.
There's all these efforts to enhance the product catalog, and that's where as an industry we've really fallen down is our product catalog is just. Total garbage and yeah, I know you guys touched that a little bit. As you know, in your world of retail media network, you're trying to make all these really cool product recommendations and the data's just not there.
Maybe you know, a user's, you know, uh, a 34 waste and looking for a bell bottom gene or something. I don't, I don't know. You just can't really pull that out of the data these days, which seems like it should be easy, but it's not. Yeah.
James Avery: You [00:19:00] see this on Amazon all the time when you're looking at products on Amazon and the only specifications are in the literal title of the product and there's zero on the rest of the detail.
And that's, you know, the most successful marketplace in the world and you still have that problem.
Scot Wingo: And then if you use the filter nav, because no one has anything, you just exclude all products and you end up with like two weird things that are not even what you know in the category you're looking for.
So we need, I would say we as a retail industry need to elevate our game there. And you know. I don't know exactly how to solve that, but I think we need to rethink the ad networks and make 'em more additive than subtractive. So one of my problems with Amazon is their ad revenue is right at a run rate of 60 billion.
It started to really negatively impact the user experience to the point where there's so many sponsored products. It almost is like. Uh, you remember that old meme where you were trying to hit the cancel button and it would move? Yeah. It's kinda like I'm looking for a product and they buried it in ads.
I feel like it has really [00:20:00] hurt the discovery process. So, you know, I think they're an extreme example. As I look at other RMN implementations, I think they're much more like back when Amazon was smaller and more helpful, you know, it is helpful to surface an occasional thing. Um, but it's definitely gotten to the point where it's bad and, you know, I think.
Uh, it's interesting that I've listened to everything that Sam Altman has said about this, and he's been very specific. They don't want to do. Traditional ads. And when he says that they don't wanna do Google, they don't wanna do like 10 blue links. And then they also don't want to do kind of sponsored listings.
But he says, you know, the Instagram ads where you're kind of in reels and you see it and it's kind of additive and it's like, like it may know I'm an NC State fan and I'm doing something on Instagram, which I don't do, but pretend I do and it's gonna show me something there. This happens to my daughters all the time where it kind of knows they like beauty products and it'll surface something in there and they'll buy it or there's an influencer they follow, right?
So tiktoks the best at this influencer thing and [00:21:00] Instagram's kind of a fast follower. So I think that's additive 'cause it surfaces super relevant things to you. So I think you guys in RMN. Should get better at that. But I think to get better at that, we're all gonna have to fix the pro. It always comes back to the product catalog in my world.
And then he talks about Instacart and it's not lost on me that the Cee o of the consumer chat, GBT is Fiji Sumo who was the, uh, she was the Cee O of Instacart. Yeah. Um, the guy running perplexity shopping is, uh, he was the chief product officer. At Instacart. So there's Instacart, DNA in all these companies, which is important to know.
And as you know, Instacart would be losing money or very close to break even if it weren't for their ad network, right? So they, they've built a good ad, and in that one the ads are in kind of like they're between find and buy. So, you know, there's up at research, which is like discovery oriented kind of advertising.
And then as you get down the consideration set. I like the Instacart ones where it's like, you have two of these Quaker oat things in here. If you put one of these other products, we'll give you a $20 coupon. And it's kinda like, you know, you do the math on that and you're like, well I'll, [00:22:00] I'll take an extra box of granola bars, you know, 'cause of the coupon.
I'm basically getting 'em for $2 and they wear six, you know, so I'll, I'll do that all day long. So I think there's ways. To do this that are additive to the user experience. But I think the prerequisite is the user experience is gonna have to elevate, uh, because the perplexity and chat BT shopping experience is just so good.
James Avery: I mean, it's interesting to play out like the, you know, maybe for the US on the retail media side, it's a dystopian view, right? But you, you get to a point if all shopping is done through. A-A-G-P-T of some type. Then essentially you are, you know, at some point you're just fulfillment. Right? It's basically almost like they become the marketplace where you're saying, I want a weightlifting bench.
And they just say, these are your options and we'll negotiate you a better deal too. Right? Like we haven't gotten into where agents could go and, you know, almost Priceline negotiate for you, right? Like go in and say, oh, well we're [00:23:00] gonna send this out and say this is the model we want. Who can get this to James in three days for the lowest price?
Scot Wingo: Yeah. I think eventually we'll get there, but you know, I think it's such a better shopping experience already. We don't have to get there for it to be a better shopping experience. Right. They'll do those things, but it is already, you know. And, and another thing is I find myself arguing with people that have never used it before.
Uh, and then also there's a set of people that use, there used to be a product called Operator that is now called agent. And a lot of people think that's what I'm talking about. That's a terrible experience. That's not what I'm talking about. Like the inside of chat. TPT, you can buy stuff, not some separate operator or agent thing.
The, the, the thing that, you know, I come from the marketplace world and what's interesting is in the Asian countries, um, marketplaces are 95% and people don't really have, Nike has a website in China. No one goes to it. Because their Alibaba store, it's actually a subsidiary called Tmall. Their Tmall store is basically the Nike store.
So people skipped websites and went right to marketplaces. So that's an environment where they have it, and then [00:24:00] what they do is they open up the networks. So for people to do more in there now, you know. That limits the number of publishers, but at least the advertisers can kind of get in there. But here in the US the other thing to remember is e-commerce is only like 16% of overall retail.
Right? Um, and then I would say, and half of e-commerce is marketplaces largely driven by Amazon, just 'cause they're so huge. But even Walmart's a considerable size, eBay's big, so on and so forth. So we're already at 8% of retail is not marketplace today. Now I think we'll lose, I do think. So my purpose in pointing that out is these things have a faster terminal rate than you would think they were.
Right? I say that, but then there's always this asterisk that I've never seen anything grow this fast. And if you see the one chart where they've grown exponentially fast compared to any other, they're the fastest thing to a billion users and by like four years ahead of everybody else. I don't know if that's gonna continue or not.
Yeah. There is a dystopian future scenario in there.
James Avery: Yeah. But
Scot Wingo: I
James Avery: do think the point about [00:25:00] e-commerce is very valid, right? Because we both lived through the first.com boom where it was, everything's gonna be online, right? Walmarts are gonna be ghost towns because everybody's just getting stuff shipped to their house, which is definitely not the world we've seen.
Right? I was at the mall this weekend, and then the mall was packed, right? Mm-hmm. And so I think there's still very much a. Physical shopping component. And I think we see this actually in retail media now where lots of retailers are. I think the way they hedge this a little bit is say, how do we focus more in store?
You go into Kroger, the vast majority of Kroger's bought in store, how much in store do they have versus a focus on digital. Right? And I think that we'll see more and more of that because that will be protected to some extent from the changes coming here.
Scot Wingo: Yeah. Uh, I do think also another thing to think about is, um, the other unique thing about the chat gt incident checkout, it's the first time.
The, the third party seller, um, and let's say Walmart is actually the merchant of record, so they're acquiring that customer and they can remarket to 'em. So I think that's why [00:26:00] Walmart's doing it. I think if they look, they can kind of see, I'm sure they did a test and they see it's incrementally new.
Potential customers for their online efforts. Um, so that it's just a customer acquisition channel for them. And I think they're seeing it as we'll get, you know, they'll come and visit us. Maybe they'll have their first Walmart experience. It's also way they've always been second fiddle to Amazon, so it's a way for them to out maneuver for Amazon a bit.
Um, so Amazon is blocking all bots still. They're, they're kind of taking the. We're an island and we're not gonna let anyone in onto our island. We're a fortress kind of model. Um, so it's gonna be interesting to see because the other positive thing for RMN is like, can you guys play an important role as they look at turning these customers into lifetime value?
Customers selling 'em more and turning 'em into repeat customers and you know, Walmart's gonna want to get 'em onto Walmart Plus and so on and so forth. Um, and that's gonna be true for all the other. You know, retailers and publishers, you will actually, I think you'll get new customers from this that you can remarket to that actually could drive website growth.
I [00:27:00] think one of the things to do is get 'em on the loyalty program. That's gonna be, you know, every retailer I talk to. That's a huge strategy for them. Yeah. Because they see, uh, we spent a lot of time in the beauty category and the big retailers, Sephora, Ulta, have these just like. Amazing loyalty programs.
We get all these good freebies and you know, people that love free little samples and an exclusive pallet or a drop. So there's a lot of tools in the retailer toolbox to build loyalty and get people kind of coming back to the store slash website. So I think it's probably overdone to kind of think that way, but it is.
I would worry, you know, it doesn't have to be go to zero for it to have an impact. We kinda learned that with Amazon. You know, they would focus in on a retailer like a JC Penney or something like that, and all they had to do is erode 10 to 15% and that can be cataclysmic for a lot of people.
James Avery: And what do you think about, I mean, something else I hear talked about a lot now is the, you know, marketing to the ai right, to the agent, right?
Like when you know, it's doing live research a lot of times, right? This is not something that's cached to index. Sometimes it's indexed, but [00:28:00] sometimes it's live research. You know, is there a world where we're, we're spending, you know, brands are spending money to try to market to the, the agent doing the research for the user.
Scot Wingo: I think so there are a couple interesting papers. Um, so first of all, let's just kind of like anchor on what they have. So they have training data, right? And that's usually up to 18 months old. So that's where it's just like it's ingested the whole internet and trained on it. This is actually important because for a brand or retailer, it actually will learn base level stuff.
Um, it's not gonna pull in your product catalog that way. Uh, it almost purposely it, it'll train on it, but it's not gonna care about the price or anything like that. Um, so that's one set of data, but that's kinda like in its brain, it's not in a database, right? It doesn't have pure retrieval of it. Then it builds a web index and that is, it's just a very Google-esque index.
And in fact, it's interesting. So open AI is a combination of being in something else and then they have their own, their building and then like, uh, uh, there's this company called Brave that licenses out one and that's perplexed to uses that and a piece of [00:29:00] something else, so on and so forth. Um, and Google obviously uses the Google search index and, uh.
So that's another part. And then it could go do real time stuff. And then now they're building a product database. That's the part that's kind of new and that's where these product cards come from. So as it's doing the web crawling, it's taking any product data, it probably keeps it in, you know, a lot of it in the index just as a, you know, here's about the products and then like, it builds a more robust structured schema ish data in, in.
In this other database. Um, so, so those, that's kind of what it's pulling from, forming its knowledge. And then the research phase, it's mostly going through training. That's kind of how it knows. Like I was, I was doing this Swiss watch experiment and it kind of knew to ask what kind of band I wanted. And that's kinda like the training data is where that comes from.
And then as it came from, and it knew a list of brands that was more of the web indexing and then it got into the product database. Um, so. You know, those are the ways you can influence is kind of those three buckets. Um, and what we do is we really focus on that product data because that's gonna really hit this.
It's harder to [00:30:00] change. The research data has shown, it's pulling a lot of that out of review content. And it gets a lot of it out of, uh, it gets general content. Like if it's gonna try to learn about Vel and James Avery, it would look in Wikipedia and if you have a page, it's gonna really kind of almost splat back that.
Uh, but if you, um, you know, where you can really influence, is it on Reddit? And, you know, there's these companies that, and there's a big battle over what to call this, these definitional things are hilarious. So it used to be called, um, uh, you know, GEO, so gen AI Engine Optimization. I prefer answer engine optimization.
I think that's kind of more appropriate. Um, and there's parts of that I think are okay. Uh, like there's, there's like a hundred companies now in this space, and they'll, oh yeah, they'll kind of, they'll run prompts and like tell you, it's called sentiment analysis in my world. Um, so they'll say. What do people feel about, uh, uh, you know, Kleenex?
Uh, what do they like and not like about it? And the problem with that, we've had that forever. The sentiment analysis. The problem is when I talk to customers, it's super not [00:31:00] actionable, but that's where a lot of the research content's coming from now, some of the geo companies, they then say, we're gonna help you build content to fool the engine.
And I think that is super dangerous. Yeah. Do not do that. There were, there were like two companies in e-commerce and they realized. Google just gave this huge page rank boost to anyone whose domain matched the keyword. So they would literally go open 600 stores and they had like hammock.com, battery powered lantern.com, all these very specific things.
They were just crushing it. Yeah. And then. Google. Basically what they'll do is they said, take notice. We think your intent if you're doing this is bad and you have X days, and then you got one more warning. And then after that they did what was called manual intervention, which means an engineer went and deleted you from the search index for life.
Yeah. Yeah. And that was a death sentence. You were done. So I've met many e-commerce companies that pushed this gray hat. You know, there's this white, gray, black hat metaphor they use. I think it comes from the cybersecurity world, but they use it in SEO. All that content generation, I just think is a huge [00:32:00] mistake.
Um, so they actually have a case study where they went to one of these financial cards and they had 'em create a listicle. So a top five list or something like that, kind of a Buzzfeed style thing. They had 'em create a fake site, create a listicle, and they were like showing up number one for everything.
And I was just like, Ugh, that's got the short term that you're enjoying right now. But. In fact, Sam Altman is basically, he said on a recent A 16 Z podcast. They know that's going on and they're working on it. Yeah. So I think those days are numbered. I would not do that. So
James Avery: there's
Scot Wingo: a whole, there's a whole cottage industry around that.
James Avery: Yeah. But the way you kind of described it is interesting, right? 'cause I think it's kind of, if you take like Nike, right? The, the kind of training data, the historic training data, you know. The, the, the engine, the answer engine understands that Nike, you know, makes fashion sneakers and they're very well regarded and that, you know, it's, that knows they have golf line.
Just like we have a brand, like We Idea have a brand, right? Like we, we have a sentiment about Nike. We both think something about Nike after 20 years, it's you and I seeing it, right? I'm not
Scot Wingo: a Nike expert, but I could kind of tell you generally what they do and how they're regarded. Just do it like we all know that.
James Avery: And then it's kind of like the [00:33:00] next phase. The next level is like, okay, where can you buy Nike's? Right? You can buy 'em from nike.com, you can buy it from Dick's, you can get it from Walmart, you can get it from Amazon, wherever, all the places you can get 'em. Yep. And then the product feed really feels like where there is the opportunity of, okay, you're dick's and you want more people ordering Nike's from you, as opposed to from Walmart or Amazon or nike.com.
That's actually your last point of influences. How up to date is your product catalog? How correct is your data? How aggressive are you on the price that you're sharing? With the GPT versus what you're listing on your website, right? Like all of those things are then dynamic at that point.
Scot Wingo: Yeah, and since I've been on this journey for a year, in the early days we would talk to folks and it was usually the SEO people at a retailer, and they'd say, there's this one Reddit review from like 2010 that I think is hurting our sentiment.
What do I do? I'm like, you can't delete it off Reddit. You know, you can't do that. You can't flood the zone on Reddit with. Positive reviews, though moderators will just kick you off so there's nothing you can do. And then they get frustrated and now we get people that say, you know, and it's funny, like it's at the Cee [00:34:00] O and board level where someone on the board or the CEO is searching for these things.
I think the fact that checkout exists is now everyone's going into that product card experience and they never really kind of rattled into it before. And they're kind of saying. None of my stuff is showing up why, and now we have answers for that. We know in that find thing, it's totally influenceable in a very safe way and it's, it's largely happening.
I have this like nine steps of optimizing this, but just kind of boil it down. There's a visibility layer. A lot of times we're finding many of these larger sites use infrastructure that are CDNs or like CloudFlare. To protect from denial of service attacks and all those things that happen. By default, those things are now blocking most of these bots, which is bad.
It's like you wouldn't block a Google bot. Everyone knows not to block that one 'cause it's free traffic, but they're blocking GPT bot and there's like 20 of these you should let in. So we actually find, and we'll find, the CO doesn't know, and they'll look at the person next to 'em, are we blocking those?
And they're like, finally the IT guy's like, yeah, you know, I thought they looked bad. I didn't know what they were. And they're like, oh my god, you guys. [00:35:00] So then we gotta get that off. And then there's site on, there's kind of SEO ish metadata that you need to kind of have to really make this work. So the crawler.
Can find your things. When the crawler comes in, it ignores JavaScript. Well, everyone in e-commerce land has all these little fancy doodads now that are all JavaScript and it can't see your images, it can't see most reviews. If you've got one of the fancy review widgets, you can't see 'em. That's a problem.
If you have a q and a, there's q and a widgets, you can't see those. And you know, it's just getting the text content. So that's another thing is we gotta figure out. You can offer alternatives in what's called the schema.org style of the metadata. So that's step two. So then once we have that, then we can start making sure it's getting your product catalog right.
And that's what we spend a lot of time on. And you have to, the way they have these cards, you want your products to show up there. And they're very finicky. And that's what we're doing is we're nudging, we're changing the data very slightly to make sure it's matching what they have to make sure you're showing up and getting in the game.
So that is very influenceable and you're not really doing anything nefarious. [00:36:00] Your intent is positive. You want, you have that product, you just want it to show up. Yeah. So I mean, I think it goes back to the same what you said
James Avery: about Google, right? If what you're doing is improving the user experience, then they're gonna be happy about it.
If Google has Google shopping,
Scot Wingo: then they, yeah, we do this all the time on there, and they're like, oh, thanks for solving that. You know, we know we're finicky. Thanks for changing your feed.
James Avery: Awesome. This has been great, Scott. Like, uh, I also, I wanted to give you a chance, like where, if people want to talk to you about this, I think there's a lot of retailers out there that, you know, are, are trying to figure out how do they get their data or their product catalog in better shape for this future of agent commerce.
Uh, where can they find you?
Scot Wingo: Yeah, so, uh, LinkedIn, I put a lot of content on there. Um, the only trick, and you can see it in my name here, Scott, with one t, that's, uh, it's a long story and uh, I also write a lot about this. And I have a podcast and a substack called Retail Gentech. So it's the word agentic, kind of jammed with retail, take the A out retail gentech.
So, um, that's where we do a lot of content on this topic and have a lot of podcasts for people in the industry. [00:37:00]
James Avery: Awesome. Well, thank you Scott. This has been, uh, I think illuminating. I think, uh, the listeners will love it, so thanks for joining me. Thanks, James. Thanks for having me. Thanks for tuning in to Unlocking Retail Media.
If you enjoyed this episode, don't forget to subscribe and share this show with your network. We'll be back soon with more insights to help you navigate the future of retail media. See you next time.