How to Get Found on ChatGPT. If Your Shopify Store isn’t “Agent Ready,” You’re About to be Invisible 🫥

Jay Myers: More than half of US consumers already use AI to research what to buy. This episode couldn't be more timely because 48 hours ago of recording this, Shopify just announced that they're bringing checkout directly into chat GPT, and not just Shopify, but Etsy, others, uh, with what is called the Agentic Commerce Protocol, a CP for short. So AI agents can complete a purchase end to end. So this is a new channel strategy for Shopify merchants where your products now really have to be agent ready to be found priced, purchased instantly. So now saying this, people listening right now, brand owners are either scared, they're may be confused, they're may be excited, maybe all the of the above. And this is why I was really excited to have Gavin join me on the show today because Gavin, you're the director at Shero Commerce. Spent over 25 years at the forefront of commerce, helping some of the biggest brands in the world solve complex problems, migrations. But most importantly for today's episode, and the reason why I wanted to have you on here is you've helped brands build strategies for AI driven commerce.

Well before I think most people even knew what AI driven commerce meant. So you've you've also become what I would say is one of the louder voices pushing brands to get agent ready for this coming wave of AI powered shopping, which again, is why I wanted to have you on the show. So thank you for joining me, and welcome.

Here

Gavin McKew: Thank you, Jay. It's really good to be here.

Jay Myers: I am.

Super.

Gavin McKew: that, that was a nice intro. That was a nice intro, sir.

Jay Myers: You know, I put a little bit of thought into those intros,

Gavin McKew: Um, um, Yeah, so I think around the agent side of E-com just to kind of give some context to the listeners, I suppose, and I. I've been a brand, I've owned probably 15, 20 brands in the last 26 years. So I've been on the side where I've had to do everything.

I've had teams who build things. I've worked for brand for that period of time as well. And I was a CTO for some big brands, and now I work on it. I work for Shero Commerce as an agency and helping other brands deliver upon that. And with that, that kind of gives a certain view of, of the things that I would be worried about if I was a brand owner.

Like what would be the things and I, but without scaring it. 'cause I certainly wouldn't be worried I actually think I'd be the exact opposite. I think this is opportunistic. It's the same thing that happened in 2004 five when you could, you know, you could run Google ads or you could or in, you know, 2008 when you could run meta ads and there's the whole period where things have to shift and the attention.

Of generalized users moves.

And it's j and all. The thing that's really happened is as everybody kind of knows, AI is big and scary. But if we look at it like at, it's really at its core without being scary, is, and I don't wanna sort of quote, you know, the old Gary V sentence here, but in the sense of like, wherever attention is kind of where you need to be.

And if attention is exactly, you know, and if attention is moving from traditional search engines to more agentic answers, which it is clearly across the board, I think attention for brands should move to that direction also. And with that in mind, if a brand, essentially a GE commerce A case of the answer being predictably given to you as opposed to you trying to find a way to hack the answer.

So if we break down what a search engine does, like just traditional search engine or even any search, I would type something into a search engine and a searching search engine has a lot of rules and it says these are, this is what your, this is the best serve results based off what you have given me.

And those rules are predated off things like, you know, all the SEO standardized things. So backlinks authority content, like it's based off all the traditional SEO services. I think a little bit as we kind of get more in, it's just the, it's just more a case of that the rules are changing to a

and it's no longer based off some of the realities of what we traditionally learned as rules.

That allowed us to be found, you know, how do brands succeed? Like how do brands succeed? Users

Jay Myers: Yeah.

Gavin McKew: then conversions. And that's it. That's that's really it. There's stuff in the middle.

Jay Myers: So let's just, I, I'm, I, we might be taking for granted that age Agent commerce is a term that a lot of people are comfortable with because I, it was, I, I remember I was at Shop Talk in March this year, and I mentioned someone asked, what is the, do they, what, what do I see as the future of commerce? And I mentioned. Agents. The word age agentic wasn't used that much, but I, shopping agents and even then people said, what? What's a shopping agent? That made no sense. And then I did a post on LinkedIn and it kind of went viral. I think it got about 300 comments on it. But the thrust of it was like, is age agentic commerce potentially Shopify's biggest threat or biggest opportunity?

And what, what does it mean? And basically everyone said like, well, that might, might happen one day. That's way in the future. That will never happen. But let's for in plain English, before we go, what is age Agentic Commerce?

Gavin McKew: Okay. So yeah, in very plain English, a gently commerce is a case of agents are not humans purchasing from brands. So making decisions to buy from brands on the behalf of humans. So there's almost, to break this down a little, if we break down three stages of where we are within the e-commerce cycle. SS stage one is I go to something, I type in my phone, I type in my browser, and I find what I'm looking for individually and then I search and I filter and find, and we've been in that stage for the last 20 years, 25 years.

Stage two is I would ask an answer engine to find me the thing that I'm looking for, but not just find me the thing that I'm looking for. It could be it has certain specifications I need to delivered on a certain time. It needs to have a certain level of reviews, like there's a certain element of things that I need to give that answer engine to give me back.

That's stage two, and we are just entering stage two right now. So we've been in stage one for 25. We're about to enter stage two, and then stage three from the agent commerce standpoint. We'll work a little bit more like this in the sense that. Your personal answer engine or your personal assistant that exists, either that's a conversational assistant or somebody you just ping in your notes could surface in any way you need will on your behalf, go out and find the very thing that you want to deliver.

And to kind of give an extreme example of that my little boys into Pokemon cards, and

Jay Myers: is mine.

Gavin McKew: and it's particularly difficult in the UK to get good Pokemon cards Now in the current way that I shop online, and this is going between stage one and stage three, but to demonstrate what that might look like, stage one, I have to constantly look at the same websites to see if I can buy those Pokemon cards.

Stage two, I will have to ping an answer engine to see if anybody has these Pokemon cards. But stage three, I will just tell my agent that I need to buy these Pokemon cards. You need to find me them, and this is how much I'm willing to pay, and it will get me those Pokemon cards when they become available.

So when we think about it from a user standpoint, my little 7-year-old j thinks that the way that I shop is bananas. Like I abs I love seeing the world in the in the sense of a child's eyes because he literally looks at the way that I, dad. Why do people do this? Just really, why do, why when you want to buy a microwave, do you have to type in that you wanna microwave and then you have to go to a website and filter 500 different things and then see if it's in stock and then change website.

I'm not saying websites are dead by any means, but the very fact that a 7-year-old is. Kind of like why do you operate in that way? I, it doesn't make any sense to me. It's kind of the very reason that we're kind of heading to that place. I suppose

Jay Myers: I have my, my son is eight years old. He asks the same questions and he is a Pokemon fanatic, so I feel like we're,

Gavin McKew: we need, they need they need to do some trades. J do you

Jay Myers: Oh, he would love that. He's got spreadsheets. He keeps track of all his

Gavin McKew: Or

Jay Myers: all that.

Yeah.

Gavin McKew: He's in, he's involved.

Jay Myers: yes, he is so, so the, okay, so what was announced two days ago was the chat GPTs, opening eyes a agentic commerce protocol, which basically now you can. Search for a product. You can say, I'm looking for a pair of shoes under a hundred dollars that are white. Something that can be here by Friday. It can find that,

and I can complete a checkout. This hasn't rolled out yet. I think it's rolling out over the next couple weeks, and not in every country, but it's essentially here. But I'll be able to complete a checkout in chat, GPT without ever leaving chat. GPT.

Gavin McKew: Exactly. Yeah. So,

Jay Myers: so, there's a lot of,

Gavin McKew: there's a lot

Jay Myers: a lot of questions. Yes, there is a lot of questions. I mean, I actually I compiled, a list of all of them, I mean, and kind of

Oh, I had,

Gavin McKew: Just the SEP questions as well. That's good.

Jay Myers: exactly. Yeah. I mean, a, first of all, for our audience is all Shopify merchants. It's the Shopify 1% show. So I think what should know off the top is if you're using. Shopify payments, which is built on top of Stripe by default. You will have that, and I don't know if it's rolling out for every country, but that's the premise, correct?

Gavin McKew: That's true. That's a hundred percent true. And the other benefit of that, I suppose as well is that you have you have catalog API as well which will, so there's two sides.

Jay Myers: Right.

Gavin McKew: You are exactly right. So if anybody is on Shopify you are currently, when that rolls out, there's a, there's an initial signup period where they're going to use signup only just to roll it out.

They don't, they can't do it across every brand. And you can log onto Shopify to do that. And you can find where that li lives. Gm I share that link on on, on either other things to do that. But the reality is very shortly after everyone will be enrolled anyway. And this, and there's kind of two sides to this.

There was two things that were announced in the last five months that matter on this exact thing on SEP. And the first thing that was announced when I was at editions dev in Toronto, which is Shopify's, you know, annual conference around development and what they're gonna build in their roadmap. I was in a, I had a chat with Vanessa Lee, who's the head of product of Shopify around this, and they announced that they're gonna be using a thing called catalog, API and a catalog, API, just to really break down what that means in its simple terms.

It basically means every piece of product data on a Shopify store is given a way through that API and certain LLMs use that API. So GPT uses it, prop perpetuity uses it in a variety of others, and it's gonna roll out. So if you update product data on your Shopify store, it essentially rules out to those in real time.

It's not like a delayed reaction. Think of it like a product feed you might do to ads, but live, right? So I add a new product to my Shopify store. Is that gonna feature in an element of whatever a an answer engine knows? I each RTPT. That's what it'll do. So I think around that was like the first meaningful thing that happened, but it didn't really have any benefits.

'cause what's the point of service and products And the way that it worked, it's worked for the last five months, is you click through to your Shopify store and you purchase on the store.

Jay Myers: Yeah.

Gavin McKew: And others, other brands have this, like other platforms. Are there elements of things you can do this generally now, but this is specifically for Shopify.

So you've f your products got fed into these answer engines. So if you said, I need a, you know, I'm looking for a new vase that's 32 centimeters high and it's a floor vase and it could, you know, it's made outta this particular glass. GPT would try to give you the best answer based off what it was being fed by Shopify.

You don't need to do anything that's happened. But the thing that was announced for eight hours ago is. Now you don't need to leave GPT itself for the purchase. So traditionally you'd be pinged off to the website. You'd then have to complete checkout as you would now you can just us first, but you can complete checkout directly in the LM.

And this creates a very interesting segment of what that might look like and how that works. What control the brands have, what control does that look like? How can we improve being more featured? Where are the benefits? And it opens up they announced that they were gonna do this when they announced they were doing the feed.

It's the very reason they did the feed.

Jay Myers: sure.

Gavin McKew: But with anything, and this is, I don't wanna get too technical, but with anything that's new, there needs to be new rules and new protocols. An SCP is fundamentally that exact thing. It's a protocol of, it's a set of rules of how you give me some data from one pro, from one system to another.

That's all it is. And when that rolls out that allows those roles to happen. The great thing about it is that from a brand perspective, you don't really have to do anything from that side. It's already happening. You don't have to worry about the protocol.

Jay Myers: So at first, this sounds great because you think, wow, I'm gonna, you know, normally when someone searches for a product through chat, GPT clicks on your site, we know that. Over 95% of them won't complete the purchase. There's, they abandon the checkout. They don't make it through purchase for whatever reason.

You know, industry average is maybe just under 3%.

So now you think, oh, great, if I am in chat, GPT and I search for something and I can complete the purchase right there. That eliminates all of that potential loss in the leaky buckets of a website. But then the more you think about it, you start to think, well, what if you don't that, you know, that example of that vase or the, the black shoe under a hundred dollars, if there's a thousand stores on Shopify selling a black shoe under a hundred dollars how, which one gets shown?

Gavin McKew: Yes.

Jay Myers: Which I guess leads into the next, yes.

Gavin McKew: Yeah there's two points on that is whether there's a misunderstanding by brands of why conversion rate doesn't happen, like just collectively. I don't mean AI wise is 80% of shoppers would never buy it in the first place, but that's just a rule. So you didn't have the 80.

Jay Myers: Yeah,

Gavin McKew: Ever. So, so you're using a full benchmark of what it's pr principally on conversion rates. I think it's based on if there's an intent to buy, so if it's hobby based or if it's needed to buy. So if you need to buy toilet roll or you need to buy groceries, the conversion rate will be higher than it would be if you need to buy, you know, saunas.

Right. It just will, 'cause there's a difference between the need and the product. So, so the, there's a collective thing of conversion rates based off the factual amount of users that is collectively done. And I think that's a slightly in Yes, I get why it needs to be done because we spending fundamentally money on clicks and we'll get onto that later.

But the reason, but getting into the element of why I think WW how does a brand get shown in a more prominent position? Within the answer engines as opposed to anywhere else. And this has been said many times, and you might have heard this, but it's, it is really about data. So if we break down what an answer engine is or an LLM is at its true core, it is its job is to predict what the next word or element or thing should be around what that is.

And if you break that down into its core of how it's trying to solve the problems of its users, it's trying to make sure that the product, that it's, the product that it's served within itself is the best solution to the question that's been answered by its user. And with that in mind, there's a little bit that's went on in e-com, I think in the last.

30 years. And that is kind of like how we kind of, we've almost programmed our brains for SEO. Like we've almost done it that way. Like to, and some brands, if you're an SEO orientated company, some brands aren't. There's sometimes an ad driven company. Sometimes they're a, you know, they generate traffic in other ways.

And it's not really a game plan of thoses. But the brands that are SEO driven companies have kind of created their own rules around how that happens. And those rules are correct, you know, authority backlinks content, like it all matters and it still matters, by the way, in it, it still matters in the agent era.

But the thing that the agentic is trying to fix and will try to solve the most is that if it's found the best answer, it's not whether I've found the best match and they're not the same thing. So if I've give you the best match, I pre get that off a load of, if you think of SEO variables within a brand, so if somebody looks at the, these are the 200 metrics that decide if a brand ranks number one on Google, right?

These are the metrics. You've got more backlinks, you've got more authority, you've got better content. You've ticked all the boxes in, like you do this, right? The, and these are just essentially weights. These are essentially weights that live within that decide if this is more important than nothing.

And around gente commerce, the weights are around answering the question of its user and fixing the query. But that will predict if you answer the question, its user and they are not. So the weights are not fixed. So weights within SEO are fixed until there's a new update on the SEO model. So Google has an update.

I decide I'm gonna re-rank my index, and then weights did change. That's like currently how it's worked forever, right? So I'm gonna put more weight on the fact that you have alt text. I'm gonna put more weight on the fact that you have these, you only have one H one tag. I'm gonna put more weights on whatever, right?

You have more backlinks and they alter the weights within the rules. A gen won't work that way. It will work like this. So the amount of data that you can give on Shopify, specifically in the clearest form of data to answer the questions that you think your user might answer is probably the single most important thing around what you can do to be ranked.

So when people are writing product descriptions. They should be written in terms of lots of technical detail around the products, answering things that you maybe don't think matters, but potentially might. So, you know, things that the user might not ask but might make a decision. The answer engine itself might make a decision whether to serve as opposed to the word itself.

Let's use this hypothetically. Let's say that you're selling fishing rods, okay? And I need to buy some, I need to buy some hooks for that fishing rod, and I need to catch some mackerel.

Jay Myers: Yep.

Gavin McKew: a code, right? So I, I don't know anything about fishing and I'm starting out, and traditionally I'd go to Google and I'd type in, I need some mackerel hooks and

Jay Myers: is the best macro hook?

Gavin McKew: What's the best mackerel? And the way that brands would've thought about that would've been is we'll mention the word mackerel and we'll mention the word hook and we'll write a little piece on this. And that's fine, by the way. That's all great. That works. Right,

Jay Myers: And you might have a blog post with the title, what is the best Macro hook that links to the product? And we've got a whole Yep.

Gavin McKew: exactly. It's exactly that type of thing in the sense of, yeah, you might have a blog post around that. And whereas now the job isn't is to say is to, I think from an LLM perspective is to make sure the LLM understands in context itself what it is that the product can serve before that gets delivered.

So what I mean by that is think about your brand. And think that there's somebody else in the room that you need to explain what that product does and if they can understand what that is, as opposed to get the word in what that is, and then that can translate the context, then that is the best solution of what that is.

So I'm probably not articulate myself particularly well there, Jay.

Jay Myers: No, I get it. I get it. We it. How do I put this? You can tell when you go to a product page that is. SEO optimized, you know, and we talk about this sometimes a lot. It's are you writing for a human or are you writing for Google? And you wanna try to write a little bit of a blend where you keyword stuff a little bit, but you still make it sound

natural.

But it's all about, it's all about keywords, phrases, inbound, outbound links. With LLMs, it's about giving all the data they need to fully understand like a 360 degree view of that product of its measurements. Every weight, every detail, how it's used.

I'm, as you were talking, I was thinking probably FAQ sections on a product page are more relevant than ever, like with

20 questions,

Gavin McKew: Well, f well, f fa FAQs are huge, right? And they will be. And Shopify brought out an app called Knowledge Base, which is fundamentally just an FAQ page, right. That exists. It doesn't exist at product level, but it allows you to feed that data. So it's really great. If anyone hasn't tried Knowledge Base and has it on Shopify, I would strongly suggest putting it on.

It's a little bit manual and it's kind of, its first infancy. I'm sure it will get more adapted into certain things as goes around that point though. think there's a big shift between brand types as well, and let me explain, right? There's legacy brands. And let's use a legacy brand as a brand that's more than, I dunno, 10 years old, right?

Let's just use that hypothetically. Legacy brand. 10-year-old

Jay Myers: Yeah.

Gavin McKew: started off on n on SEO, fell into the influencer marketing side, chucked into ads, handles it in whatever context it does. We are now into a point where I think there's an opportunity for more disruption for non legacy brands. If, I think, if there's any level of concern, it should be around the legacy brands pivoting to make sure that they are doing this and not resting on their laurels.

The fact that, you know, we spend XY you know, we spend 45% of our revenue on ads, so therefore it converts. Whereas the users have all left where the ads live and the ads then, you know, they might get cheaper, but it's a market 'cause no's using them a around as opposed to the kind of distributor brands.

I think. I think AI and the Gentech opens up huge opportunities for anybody out there that's even considering about starting a brand. To not think about product pages as product pages to think about them in its in itself. I won't say the word manifesto 'cause I think that's crazy but I will say in the sense of is if you could distribute it at that level where you can alter how that lives within the answers, it could be delivered.

I think that a lot of dis disputed brands could really shake that up. If you look at the go-to plan for brands in the last six, seven years, right? Is start a brand, have some money. Fundamentally buy a product at relative at 25% cogs, like I speak very frankly, and spend 45% on marketing and scale it, right?

That's like the literal strategy of where that lives and and listen. Strategies work and that works, right? That's just what, it's 25 on cogs, 40 on ads, chuck it out, get traffic, pull it through. Okay, that's great. That isn't gonna change overnight by the way, guys. That's not gonna happen.

People are still gonna use Facebook and Instagram and people are still gonna use, you know, Google to find things. I think that, I think it's the percentages of the marketing side that's switches significantly. And the thing that Shopify brands can do the most is there's two real things. I think the first one J, is your product listing can no longer be weak, right?

Brands that have decided they're a very ads driven platform, that it doesn't really matter if their product descriptions are strong. I think that's gone. I think it at least in the description cell on Shopify, add some level of content. There's no excuses anymore. It's got you can literally, or you can write two sentences and get it to write four paragraphs.

So just get it to do that. The second thing I'll say, and I think this is a bit of a hack, or not a hack, but like a thing that's very important is, I know it in catalog API, that it re, it feeds off every single meta field that lives on a product, right? So if you make a meta field that lives against the product,

You can, you automatically know that's been served through catalog API and therefore if it's automatically been served by catalog API, it's automatically going into

From a brand perspective if there is information that you would like to feed in to an LLMI. My suggest let's say a product level could be a collection level, could be a brand level. My suggestion would be is make it a native meta field on Shopify. So make a native meta field, give it the context that it needs within that meta field.

So hypothetically you'd, let's use the fishing rods example I used earlier, Jay. What is you know, what is, what are these hooks good for?

Jay Myers: yeah. Fishes species. Species can, it can catch

labeled that as the metal field and then put the fish species.

Gavin McKew: And exactly and list them separately and say, these species this is what it is. And write that out verbatim. So it is very clinical that the, that, when that gets into the LLM and it gets its learning data

Jay Myers: Yeah. That's

great. That's great advice.

Gavin McKew: and two of the, sorry.

Jay Myers: eventually, when every brand starts doing that though, it's kind of like I, I see it as a little bit like some of the early days of SEO. There's things you can do like this that are like easy wins now, but the LLMs are like, the cream always has to rise to the top, right?

Gavin McKew: It will the cream will rise to the top, I think, through a lot of other things as well. And it is around data. All of it. Reviews are gonna play a more prominent position than ever before in life, ever. Right. So the human element of this, so the human part that when we do a purchase, we are willing to say that this brand did a great job, is gonna play a drastically bigger importance in the non-human part than anybody re really kind of can compute at this point.

I don't think so. Reviews.

They surface on LLMs. Like right now we're currently building protocols to make it work. We're just not it's the equivalent of I've made the protocol to make this tap run hot water. Right? And then I'm going, yeah, but I need to swim and win the Olympics.

And I'm like no, but you haven't even put, you can't even get a whole war yet. So just so there's kind of, there's, we're there's stages to this, to a degree.

Jay Myers: Yeah.

Gavin McKew: We're on the, we're on the protocol of building this out stage. But be sure to make sure that you're on review platform that is thinking about this speak to whoever works for your review platform to make sure that they're considering, because this is not gonna judge, it's not the data itself how much the L-M-L-L-L-M has the, has of data to give it the best answer to its user.

Is go is literally gonna be the metric. What I mean by that is, is I, what did I buy earlier? I bought I bought some top loaders for the Pokemon cards earlier. Right? And I went to Amazon and I clicked on the reviews and I read 20 reviews on some top loaders. 'cause my little boy doesn't like particular ones.

But in an agent world I would write that. I don't want that to be the case. Now, the product listing would not be expected to have that. These weren't, that this particular product didn't do a particular thing. But the reviews themselves will dictate if they do a particular thing. So he doesn't like a blue tinge in top loaders.

That's it. I don't want any blue tinge in my Pokemon top loaders. Okay Dad. So I look at the product, I'm not reading the product description. 'cause there's no brand telling me

Jay Myers: No blue

Gavin McKew: That there's no blue tinge. So I look at the reviews and I immediately see that this brand says, by the way, there's no blue and I buy it, right?

Which is great, 'cause like, why did I have to go to the product page? Why did I have to read the reviews to do this? But the answer existed in the reviews. The answer I wanted existed in the review as opposed to the answer existed in the context of what a brand can deliver. So, right now we're not there yet, so we can't do anything on that unless somebody wants to come up with the biggest hack in history, start making meta fields for every review.

But let's not go there for now.

Jay Myers: Well, I could see some of the reviews apps have having their own metadata

Gavin McKew: Oh, yeah. Well, of co they will do. Yeah, you're absolutely right.

Jay Myers: But that won't necessarily feed into

Shopify catalog.

Gavin McKew: Well, well, it would but you have to think about it as three different layers. So Shopify as a platform is the platform, right? That's where it lives because it's a big platform. The biggest eCom as platform, you know, in the world.

It has inroads to get your data into these platforms first. That's a great benefit. things that live around that, IE your reviews platform right, are not there yet, right? So they're unsure of themselves of whether they feed these through a CP directly into the LLM, which is probably how it will work from my CTO brand, but that's probably how this will work.

So you'll probably find, you know, the big guys, you trust pilots of the world and your fee fours or feed directly into the LLM mapping and what that looks like. That also needs to be a constitution of of how. That product data, kind of how does it match and know that's the same product. So if there's a single product with a single SKU that's collectively stored by you know, 500 stores worldwide, the LL m's gonna want to know that's stored by 500 stores, right?

It's gonna matter. Whereas in, in SEO at Walmart it like, it, it will only matter what's on the page that you visit. 'cause it's judged off backlinks and contexts. Whereas the LLM will fully understand that if you are buying an iPhone 17 Pro 128 gig in black, that there's all these other stores that sell this product.

There's loads of them, right? And reviews will become a little bit non universal to a point as well. So, because other people will sell the same product and it will re recognize that it has the same IBIN number, it's the same piece of data, and I can solve the same problem from someone else. Now, the weights that the individual LLMs put on what it actually surfaces on its own platform are yet to be decided and yet to be

Jay Myers: Yeah.

Gavin McKew: So right now they're data hungry,

Jay Myers: Yeah.

Gavin McKew: so feed them is the job.

Jay Myers: So on the reviews, would this be a good strategy? I know a lot of times when brands ask you for a review, they have, I don't know, some type of incentive, like you get a 20% coupon or something else. But if you specifically, if I'm a brand, knowing the importance that LLMs are finding data about the products to answer questions instead of the reviews being talking just about how much they love it.

Great product. Oh, my OMG. I love this. Ask customers to say why they

love it.

Gavin McKew: I think that's a great jab. I think that's a great solution in the sense of I think from a brand perspective, the more context that can be put on and less of the top level, like top purchase was on prime type context that we see in, you know, you know, look at eBay reviews.

It's almost a joke. Like it's just a like I have to skip through. Now

Jay Myers: Plus.

Gavin McKew: The, that, that's not what reviews are anymore, right? Reviews are like, reviews are, tell me why this was great. Tell me why this was great, and I'll give you 20% off my next purchase is an absolute fantastic angle. And actually. In the future will be more valuable than any type of than a lot of types of contextual, you know, free gifts, discounts that live on the PDP.

And certain market hacks. Brands will be craving get getting context into reviews around user sides. Obviously that means that they're a good brand. The ones that are good brands, but the ones that aren't, you know, sell a good product

Jay Myers: I mean, if I'm a reviews app listening right now, I would be changing my input fields rather than just click one to five stars and leave a comment to three questions. What do you love most about this product? Why? Why did you decide to buy it,

and what would you, and what would you, yeah.

Gavin McKew: hours a go. Gaja. This was not a conversa like, and that, that's what I love about this in the sense that's why this is the most interesting time I was speaking to her, and I won't say the name. Probably, but what probably one of the biggest search apps in the world this morning. And somebody their product team and around the call we, I was having at work and I think they were talking about how search is gonna be affected.

And I said, but think about it. Think about search in the context of either what happens on the site or doesn't. If you sell search, it's your fundamental job, but if you sell search and you are selling it to an LLM you suddenly, you're not selling it to a user. I'm selling it to an LLM, right? So if you sell a product, I sell an iPhone, right?

I sell this for, and my job's to sell search. Traditionally how I would sell search would be, I'm looking for an iPhone, I'm looking for a black one, this gig, whatever. That's how I'd sell search. But if you sell search in the future, you're selling it through a CP

And the context that lives behind that.

In order for that to understand why they should buy that from you. And your job is to add more context as the purpose and the, and the variables as to why that happens. So whilst, if you look at this through a traditional lens, if anybody looks through this traditional lens, whether it's a search app, whether it's a review app, it doesn't matter in any level of application.

I won't get into big tech stack talk here, but like in the context, if you look at, if you look at it in the sense of contextually, I think this is, I don't think I can make this work. It's of course you can search is just moved and how you deliver that search is just moved. How you deliver reviews is just moved.

But from a brand standpoint right now the lucky thing I suppose about where we are is there's not a lot to do because not a lot of users have pivoted. How many people are in, in two months? What percentage of people are gonna be shopping on GPT? Probably very small, but. In three years, what percentage of people are showing on GBT, and that's what matters.

So I think, you know, as we get into the holiday season and the Black Friday season, you know, e-commerce managers, CDOs everyone in dev shouldn't necessarily hugely be panicked,

Be. They, but they should be putting a plan in place on how they can enrich, on how they can enrich answers to these machines in order to deliver the answers that are correct for the brands that they have.

Jay Myers: It's gonna be an interesting time because, I mean, you, I can't remember if it was before or during the episode, but you know, you talked about Gary Vaynerchuk talking about how there's points of time to take advantage of something and you know, when, oh, you know, it was during, when 2004 with Google Ads, I was there.

I ran a store and I remember buying. Cents a click and it was the best thing in the world

and

Gavin McKew: gi I run ads for free for two years. For two years straight. I had, I turned it up, I run Google Ads for free for two years on four brands and sold all of them fully.

Jay Myers: If only you could go back

Gavin McKew: was what

Jay Myers: But now we're living in a

new time that we'll

look back on in a similar way. Because right now, you know, I think everything you've said has been super important for brands like, okay, I mean, we can get into a lot more, but like the product page, giving it information, structuring reviews. But the where my mind goes to is, okay, in six months, are we gonna start, are we gonna start seeing, well that too, but also every product page being super long form

with tables and grids and okay, so if it does.

Gavin McKew: of course I think if we were Frank though, Jay, I think if we were absolutely frank about the big brands. Any brand doing more than, I dunno, five mil, right? Any brand, right. Worldwide. of them are spending money on ads. Right? Like a significant amount of their revenue on ads. That's just the collective case.

There's no point hiding around. Right? They, it's just, that's what, it's that any brand above 5 million pounds, dollars, whatever, right? They're spending a significant amount ads. I think the thing that does change on ads is where it's spent. That's the only thing that I'm saying. So it's a case of if brand, if there's a brand selling fishing rods today, and it's a $12 million brand.

Okay. And it makes a million dollars a month in revenue and it spends $450,000 a month on ads, $200,000 a month on staff, $200,000 a month on cogs, and it makes a hundred thousand dollars. I'm not saying that's like the formula, but I'm just demonstrating that's not unusual. And those ads are now surfaced somewhere between, I dunno, TikTok, meta and Google.

I think the percentages of those ads in the next 18 months change drastically because. Opening I is gonna have to IPO, right? So they sign a 6 billion dollar fundraising round about, I think it was about a year ago now, just under, and in that fundraising round was an agreement that they have to change from a nonprofit to a profit.

This matters, by the way. It really matters. So, and when they have to go to a profit company, that means they're gonna have to ipo. And when they ipo, they're gonna have to raise revenue very quickly. And right now, charging $20 for cha, PT Pro isn't gonna cut it. Right.

Jay Myers: Well, I think right now they're like the drug dealers getting everyone in the world addicted. If I'm on the $25 a month plan for Claude Chachi, PT Gemini, all of them.

And now if they, I lie on it and I

have

Gavin McKew: at the Amazon Prime did the same. Amazon Prime did the same though. Like you have to do that. There's an infrastructure there, there's a loss making any monster company. If you are thinking big plans whether they succeed in the long term is if you are thinking big plans, there's a loss make in five years.

There just is, there's an infrastructure layer that needs built. Like they just always will. 10 years, you know, when Amazon built the infrastructure layer, but it was buying all its warehouses and its vans and building its distribution layer. That's of course what happens. That's that's not insane.

That's what you need to do. But from the ch, from the from the GPT standpoint though, because they'll need to IPO and I think this will happen. I don't think it happened be, I mean I don't think it'll happen before this Black Friday. I think they would love it to, but when that happens, there's gonna be this very immediate switch in e-commerce world, right?

So, I dunno, let's suppose that your average cost per click on meta and on Google is a dollar hypothetically, right? And your conversion rates 3% in any niche, and then all of a sudden GPT comes along and charges 20 cents for a click, right? And your conversion rate's 5%, right? What do you do the following month on your ad spend?

When you're sitting down with your market manager and you see a more, and you're having a conversation, you're going, right, we need to put more spend in this. Now there, there's a case of like chicken and egg here, though, are users there for the demographic that matters? You know, the demographic that's using GPT more is definitely Gen Z and under right now.

That's 30 and under a ki like they, they don't Google anything anymore, right? They just don't literally they'll just

Jay Myers: Yeah, but even when something, it's the AI results are first anyway. So you're using

Gavin McKew: They are but they're still trying to sell a click. You know what I mean? You gotta think what Google is. Google from its fundamental core is trying to sell you a click. It's not trying to sell you anything else. It's why you have to scroll for two minutes now to find things that aren't clicks. So it and from a sales standpoint and whether users understand that, I don't know.

But I look at this from a Mark marketing development standpoint and it's got it's got a little bit worse on that. I think product list ads have been pushed down significantly. Performance Max is, as across any brands I know of kind of, they're struggling to a degree. And again, it's, it is about where attention lives and where people convert.

I always have a certain thing about I'm not a ma I never was a massive meta fan in my brands, to be honest. I always saw products that were good and not trying to like, and that isn't like a hack to anyone who sells massive on meta but they were always more expensive and. I think if you are not in the impulse scenario to buy, I'm not in, I'm not in the mindset to make a purchase, then I won't buy.

That's just the way it is. Same with the App store, Jay. Do you know if I'm not looking for an upsell app, I'm not buying an upsell app. I'm just looking for one. a big difference, right? And I think there's a the reality is if somebody on an ans on an answer engine that we're now calling them IE an LLM answers asks a specific question, it's the level of specificness and then certain variables within the question that will dictate whether or not that will become a transaction.

Right. So what I mean by this is I ask a question like, I need a black microwave that's 700 watts. But if I ask, I need a black microwave that's 700 watts and I need it here by Friday, that is significantly transactional, right? So around that point, that's where I think the ads platform itself, which will exist very soon across LLMs will more lean into, so it will charge more for phrases that are more transactional and more conversional than we have previously seen on any other platform.

But to start, it'll be very cheap. It will get everybody on board. Plans will grow up. Every brand on Shopify should immediately switch, market and spend, by the way, like right away and the or. And if you want to switch market and spend, this is what I will say is get your product feed in order. If you do not get your product feed in order and get your data correct when the switch happens, it's gonna be problematic to move to those platforms.

So whatever they connect into, they could, you know, they could connect in directly. They could ask for a feed. Sorry, JI was, I interrupted you there?

Jay Myers: No, I was gonna say, I'd love to like actually go into that. So what does someone, I think there's probably a lot of people listening, nodding their heads going, okay, I'm in I, what do I gotta do? What do I like?

What are my steps like right now?

Gavin McKew: steps right now is, steps right now is export every piece of product and collection data right now. All of it, right?

Jay Myers: okay.

Gavin McKew: Take everything off site, look at it, review it, give it to your teams, ask them to enrich that data, find whatever data you can online, enrich it. I'm not trying to sell anything here, but like there, there are ways to do this.

You know, chiro has ways that we can enrich data that like automatically through AI to do that, to make it more

Jay Myers: And when you say enrich data, you mean your, like you export your product, CSV, you have all your products and now like descriptions, titles, all the attributes.

Gavin McKew: well let, well, let's use the fishing rod example. And again, I kind of go back to that in the sense of there's a fishing rod and I sell hooks, and these hooks are really good at catching C bass. Right.

Jay Myers: Yep. Yep.

Gavin McKew: And right now that's what somebody has WR out. They've got a 25,000 skew collection.

And they're a 12 month team, they have no time whatsoever to update that collection. 'cause like they could probably do one product description update every 18 minutes for a lifetime and never get through it. So what happens is they just don't bother. There's there, so they kind of forget the fact that matters, like really matter.

And, but really what they should be doing is going into their key product lines, making sure that they are on they're informing that product line. Not looking at from a Google lens, looking at it from what problem does this solve lens and writing the answer to that problem. And think about it like you are speaking to a, your 10-year-old son.

Write it out like that. So this hook can catch mackerel, it can catch code, it works on a Friday, it can work on boards, it can work on shore. Right? Do you get what I mean? Right. So, and so, so explain it in those terms like you are trained. Think about it like you are training something as opposed to giving the words to something and write it out in as much context as you can.

Certainly for your key product lines. So, big brand on shop, a brand doing five mil on Shopify now, and they have 200 products and 20 of them are their big call ads. Go into those 20 products, all of them, and write out every question a 10-year-old would ask about that product. Into the product description, add it.

If you can add it into a meta field about what that actually is answering, that's even better 'cause it fragments it out. But as a minimum added to the product description, make sure that infantry's correct, makes sure that all context that it lives on, that PDP is correct and collection page, if you do that, you'll be a thousand percent ahead of most people that even think they're ahead of this game.

Jay Myers: Yeah. What do you think happens when there's two brands that have perfectly optimized products?

Gavin McKew: Yeah. So I not, well, what, well, again, I think that's a, that is I was speaking to somebody in AI about this the day. So. AI at its cordier, and this is like a more grander question, and this is not to scale brands and merchants but AI is a predicting engine. It works principally like Grammarly does, to predict that you are gonna write something but on a much grander scale and in a much higher context window.

And the weights that live around that model, and what I mean by weights is the things that you give the highest level of decision to. So is the highest level of decision by Cha GPT going to be a. The fact that you give, is it speed? Is it the fact that you give me more data? Is it the fact that your site flow quicker?

Is it the fact that, that I have mo I have the lifest access to your data? Is it the fact that you are cheapest and I think it'll convert because I would buy that if I was you. So there's a whole horse of these new rules that are not yet fully predated. I look forward to working out what those rules are, by the way and we will very quickly.

But one thing is absolutely certain though right now is the answer engines feed on data. It's their most valuable asset. So if there is any piece of data that seems mathless, let me give an example, right? You sell these fishing rods and you don't wanna say that they're made in China because you think your audience won't want that, right?

Is. Make a meta field add that it's made in China right now, just add it just because there's some people that will like that. That's actually a positive as opposed to a negative. So you need to kind of think of, what I mean by that is, is I don't, American made goods are expensive. I'm after a cheap,

and then there's a context that lives behind that, that it's not So it's the con, it's the context.

I'm not using that in terms of frame. I'm using that in terms of it. There are many things that you can talk about any subject matter. I could talk about your baseball cap for 20 minutes, right? If I can talk about your baseball cup, 20 minutes market. It should be.

Jay Myers: I, and I think what I'm hearing from this is better to be a position than to not, right. So rather than not saying it's made in China or leaving out that it's something, there will be a segment of shoppers that, that is a plus better to show for that segment than to be vague.

In your product description with fluffy marketing,

The number one, trusted, number one, blah, blah, blah.

Be

Gavin McKew: a agree.

Jay Myers: Yeah.

Gavin McKew: Agreed. There's five words that I think are very important. And I shared this on X yesterday. And I think this is kind of like a direction of where we're about to go, right? And I opened up chatty tea and it asked me, what's on your agenda for today?

And he never asked me what I want to search for. He never asked me what I want to solve. It was trying to find out the problem that I was trying to fix,

So the very thing that I'm trying to fix now, there are plenty of shoppers. To explain on this, if you think about this from a shop perspective.

Jay Myers: Oh, my mind is going a million miles a

Gavin McKew: Yeah, exactly. Yeah, exactly. That's good. So my wife loves nothing more than going into a big shop. Loves it. To me, I walk in that shop and I'm like, and I would say it's a very, and I'm certainly not being sexist by saying this but it's certainly like lots of men I know are very much in that manner and lots of women I know are very much in that manner.

And but in the same vein of what I mean by that is, is she gets the light out of the things that she could potentially having her life. Whereas I get the light out of running over people in Pacman. Right. So, and it's, and it's a very, it's a, there's, it's a very distinctive shopper experience, right?

And websites right now are definitely serving the master element of, I like the brand experience. I want to feel what you're about. I need to feel the emotional connection. That's a hundred percent where we are. And that's not going anywhere tomorrow. But there is a mountain of shoppers that are like me, that need problems fixed, that cannot be bothered looking at websites.

Jay Myers: Yes.

Gavin McKew: Right. And the only reason Amazon killed it, it's only two reasons. It's its distribution, network and delivery meets its large catalog. That's it. That's the only two reasons Amazon kills it. It's what Amazon's PDP. It's potentially the worst PDP that's ever been invented by anybody ever. It's an embarrassment of PDBs and I don't care what anybody says, there's no a b test and that can tell me otherwise.

It's a joke. But it wins because I get my product tomorrow and I've got a massive catalog to find the items I want. And that mountain of shoppers, the disruption layer that I think Shopify brands have to take and displace people from Amazon is crazy. Like it's crazy. The dis

Jay Myers: you said something, you said, what is the problem you're trying to solve today? And that just made a light bulb go off in my mind. You've obviously probably heard of like Clayton Christensen's job to be done theory, which is like, everything does a job, right? Like

black shirt. Yes. It clothes me, but it also, uh, makes me look casual, makes me look somewhat professional, but still casual, whatever, you know, it has, does a job.

Right. And when you said that a, like a light bulb just went off because that's the way people ask. AI for things like, I'm looking for dinnerware for a romantic evening with friends or whatever,

but we don't describe products that way. But if we start thinking, describing them, the job they do, the problem they solve,

Gavin McKew: and I need it by Fri and I need it by Friday is another one. So what I mean by that is, is if your delivery time can be delivered by Friday, hypothetically, right now, there'll be a lot of Shopify brands there that are now, that is now surfacing delivery time in non SCP connected elements,

Jay Myers: Yeah.

Gavin McKew: They're just surfacing on PDP through some fractional non element is, if you can tell. If you can tell the LLM that it's delivered by Friday because you've got a nice dinner date with your old college buddy and this is what they're into, what should I wear?

Jay Myers: Yeah.

Gavin McKew: And I think as well, there's a secondary part to this as well, which I think is more interesting as well.

I in the sense of right now, the only thing that we've ever had to deal on product data is around, we've had user journey and we've had cookies. And I will leave this thought with the user's minds 'cause it will open up a massive element of where this is of where this could look. But let's suppose every user has their own agent and it knows more.

Your u your user knows more about you than your partner, than your mother. Than your son, right? It knows more about what you look for, how you work. So you're not talking, I've looked at these websites, I've buy these shoes, or I like to look at these brands. It knows those things. When your agent and when your user's, agent's buying from agent to agent, which is the third stage of the thing, we very much started with what will they need to know to buy from that agent?

So at some point in the future, I will, I won't walk in that massive department stage and be bored to my brains, but my wife will be delighted. I will just say, this is my problem. You know everything about me. Go and find another agent that can fix my problem and then transact and come back and tell me you've finished it.

We'll leave the whole conversation around what AI will look like in e-commerce in five years on that, I think.

Jay Myers: Yeah. Yeah, that's, uh, it's an exciting time and it's also, it's, uh, I think it's an opportunity. I think everyone listening should be embracing this and excited as much as possible and spending a lot of time, uh, investing in things that we're talking about today. Because if you could go back, like you talked earlier, if you could go back and do 2004 again or 2003 with

when Edwards came out you could probably be a billionaire if you could do that again, right?

Gavin McKew: I think you can do it again now though. I think if you sell if you sell supplements to I dunno, weightlifters to date. And you're not an Amazon brand, hypothetically,

Jay Myers: Yeah.

Gavin McKew: and you are the 50th biggest brand in America. Right. And I mean online, I'm not talking channels, I'm just talking like on brand store.

I think you have a ridiculous level opportunity to be the second biggest brand,

It's a case of jump in the queue is what I'm saying. It's not the fact that there's anything to worry about. It's the fact that if you look at, I sell the most amount of handbags in the us. I sell the most amount of flack caps in the us.

The opportunity lies for the hundred and 50th biggest selling flap caps variety in the US because they can definitely be the top 20 because that's the pivot. It's not the fact that the one needs to worry or the 150 necessarily needs to worry. It's the fact that these, this like level of divisions that I think brands have the opportunity to jump now.

Why be scared? People are still gonna transact. People still need products, right? They're still gonna buy 'em. It's not gonna change. It's just, if your tactic is very universal of, this is my current formula to achieve that, well, that's not gonna be the same. That's fine. Right? That's fine. The brands that are looking other ways are gonna overtake you.

So I don't worry about e-comm. I'm so excited about it. I don't worry about Shopify brands in anywhere. Like it's gonna go up. There'll be more spend on e-comm in the next two years than they'll, than they've ever been. That's a certainty. But the way they spend and the things that they need to be delivered will drastically change.

So it's the jump, it's the order of play as opposed to beam scale. So I just wanna point that out is I'm excited hugely by this Jay. Like I, I can see how this is about to play and I think everybody should get excited about it because the all guards potentially. I'm gonna have to, you know, step aside to a degree.

And I think that's exciting for Challenger brands.

Jay Myers: Well, I think this is a great way to end it. I definitely could go on for an hour or two. I think I got through a, not even a quarter of my questions, but that is, that's okay. That's how it goes. I'll, well, you know what with things changing and evolving so fast in the space, I'd love to have you on again in the future because it'll

probably

Gavin McKew: I love that.

Jay Myers: four months from

Gavin McKew: love that. It'll certainly be different in four months than now. And I think understanding that space and I think as well, kind of reassuring brands that it does change to a degree, but really right now is it's a case of get your data ready and make sure your market and spends ready.

And there's a lot of other things that live around that. And the value that you can give around what that does things will change drastically in five years. So. I think I really, I've had a really great chat with you there, Jay, and like I've super enjoyed it. I really appreciate you asking me to come on.

Jay Myers: Oh, me too. Where can people go to learn more about you if they want to connect with you? What social platform? Vonne, where do you wanna send people?

Gavin McKew: Yes. Yeah. So, so, yeah, I'm on LinkedIn. It's my most prominent channel. I don't write that much to be honest, but I work for Shero Commerce. That's like my, you know, that's what I do. So she, commerce is an agency. If anyone's looking for any advice around anything I've talked about, please do visit hero commerce and have a chat with me.

I worked on brand, so I know what it's like to work with an agency and I understand like how you can cut through where it needs to live. And that's, I'm on Twitter and you can kind of follow me on there, but LinkedIn is definitely where I write most of my content and most of my learning.

So it's just Gavin McKew on.

Jay Myers: And I'll make sure to include all those in the show notes as well too. Gavin, thank you so much. This was a ton of fun.

Gavin McKew: Thank you brother.