How Agentic AI and LLMs are Changing E-commerce

Chris: [00:00:00] Hey everybody. Welcome back to Seller Performance Solutions, Chris McKay with ecommerceChris here I'm with Matt Kostan, at ProductPinion. We've had you on a couple times. Matt, how are you doing?

Matt: I'm doing well. Thank you for having me.

Chris: Yeah, of course. As we were discussing a moment ago, the agentic AI shopping experience is all the rage.

I know you've been to a couple events maybe the last couple of months, but this must be what's discussed in most of the circles. You're moving in, and I assume that is mostly around how are buyers looking for product now versus how they were doing it six months ago, 12 months ago, and what you need to do strategy wise.

Matt: Totally. So, the game of searches is changing, obviously. Where now we have these agents that can basically go shopping for us, which is cool. And obviously that's a big push and a lot of people are talking about that. Like that's the next big thing. And I agree, like I hate having to shop [00:01:00] for mundane stuff like the usuals, the basics.

I'd rather be like, Hey, ChatGPT, get me the toilet paper. Whatever, paper towel that's the cheapest, whatever. So I think a big shift with that is, and it makes sense even in today's market too, where if you're fairly generic, if you don't have like an actual brand, like a lot of people think just slapping a logo on something is a brand.

Chris: Not anymore.

Matt: No. Because I think one, one rooted thing with people is that, people depending on the category, actually enjoy the process of, of shopping, right?

Chris: Mm-hmm.

Matt: So if it's particular things that aren't necessarily like routine, they actually do want to buy and associate with a brand because that's what gives them identity, right? A brand gives a proper brand, gives people identity. It's how the big brands are able to sell ridiculously expensive stuff because it gives people a sense of identity.

Chris: Right.

Matt: So [00:02:00] something I've been thinking about big time is, yeah, like how can brands stand out by actually implementing that and actually standing for something.

And one of the strategies that will help with the LLMs is literally if you want LLMs to mention you, give them something to talk about. So anytime you can name a specific feature or you can come up with some specific thing that your brand does so that you know when people are searching, like, Hey, what's the best whatever for whatever category, the LLM has some extra like tidbit or a little something, something that'll be like, okay, I'm gonna recommend this one because I've seen all these citations.

I've seen all these backlinks referring to this one and referring to this specific thing.

Chris: So functionality, like the bullets on your detail page would be scraped and..

Matt: Yeah. So for example, like Apple, I mean this is just a good standard practice anyway. So for example, like Apple has a retina display.

What the heck is a retina display? They made up the term retina. Right? Just sounds cool. But you [00:03:00] know, now can we influence an LLM to be like, oh yeah, which has the best display out of all of the computers? Maybe the LLM will be like, oh yeah, Mac is great because of their retina display and it's been cited this many times, that sort of thing.

So it's basically just making, in essence, it's just making things up to make your brand stand out. That's for the LLM but it's also for like a person wants to know, like when they're buying stuff, they want to feel a certain way, and you're able to kind of, as a brand, that's your mission.

It's your job to kind of create these things, to create that value. Another great way of doing this, you don't have to just make up features, name features and whatnot, which you should do, and you can trademark them and there's a whole other fun thing around that. But you can also stand out by having, a good guarantee.

Right. I feel like that's even table stakes these days where people are only gonna want to transact with you if they know that there's absolutely no, um, you know, money back guarantee is one thing. Naming your guarantee is another thing.

So for example, one of the brands I work with, it's a travel brand and, I [00:04:00] say like, lifetime never lost guarantee for the luggage tags.

That stands out and people resonate with that and they choose my product over the others because of that. And the hope and the game plan is the LLMs will use that as a differentiator when they're recommending products, when someone asks.

Chris: Yeah, I guess part of this frenzy of activity around AI and how it's going to impact shopping is.

What is a one-off or a one-time thing, and what is a new cemented experience that you can replicate, right? There must be tips and tricks people use to get like a short burst, but is that short-term thinking versus long-term thinking? Or do you start your basis of long-term research on the short-term experiments?

Matt: Yeah, I think when it comes to, I think consistency is a big play for at least optimizing for your brand and for any AI thing. So the more resources that an LLM can pull from that. Mention the same thing like, Hey, this is the best X because of y, and it can [00:05:00] cite more of those.

You're that much more likely to become the resource, to become the answer. When it has more sources. So very similar, like back in the day with SEO more backlinks, backlinks or votes. The more votes you have, the more likely you appear higher on the Google search results. Similar idea with the LLMs, but in this case you don't even need backlinks.

It's even easier where you just need people mentioning stuff. So doing Reddit marketing and other forums and things where you can have people like mention specific things over and over again of what you want your brand or your product to represent. Can go a long way for getting me more mentions in anything ai.

Chris: Yeah, I mean, you guys, ProductPinion, you're the masters of AB testing. So is this an AB testing dream, everything that's coming down the pike with ai or is it too difficult to strategize because there's too many AI tools to choose from and you have to ab test a million of them and it's like cumbersome.

Matt: Yeah. So I mean, the interesting thing with the ai, the interesting thing is like making [00:06:00] creatives and making options have never been so easy.

Chris: Yeah.

Matt: Especially options that are just good enough for testing. So obviously we need, a lot of times, we still need a human in the loop, which Danny talks about all the time, McMillan from, Seller Sessions, to fine tune what the output is. But what's exciting for us is that making variations that are just good enough for someone to run an AB test and see how their main image would perform in a simulated environment as if it was on Amazon. It's easy now. It's super easy.

And there's, specialized tools like listing optimization ai, that make it even easier. 'Cause it has the context of Amazon powering it versus just trying to go directly to like ChatGPT or Gemini or whatnot.

Yeah. It's exciting for us because it does mean that, there's just more to test.

Right, which is the name of the game. But then I think what you're alluding to as well is that it makes the process like we have to nail down more of a culture of testing and more of a culture of process, of how to actually accomplish this. 'cause [00:07:00] obviously you can get lost in the data, lost in the options, lost in the thick of things.

Chris: You might hear, okay, here's my presentation of our results. And then they're like, okay, so what choice do we make based on that? Like they might grasp what you're saying. And there isn't tons and tons of historical data from using rufuss or whatever to, to base decisions on.

I mean, it's all kind of in its infancy, right? Or do you think it's maturing now? I don't even remember when Rufuss became an everyday conversation, to be honest with you.

Matt: Yeah, like about two years ago when they launched, and then obviously now it's more, I mean, it's growing, obviously it's growing, it's becoming based.

I think it's just gonna be Amazon's default search. It's not gonna be anything, there's not gonna be a difference between search and Rufus. It's just literally gonna be on there and something that's cool that we're working on is, a third like us studying just how shoppers are actually using Rufus. And we have a unique ability to actually, watch people shop on Amazon so we can actually [00:08:00] record it. So one of the tests that we have. Yeah. So one of the tests that we have is like focus

Chris: group style kind of?

Matt: Yeah, yeah, yeah. So basically, but more scalable. So you have these individual sessions where their screens are recorded, they know they're being recorded, they're answering questions out loud.

We can even have them do certain tasks as they're being recorded. So we can literally tell them like, Hey, interact with this, Amazon's AI chat bot and then just watch them and see what they're actually doing. And then obviously, these are individual interactions, but when we put a bunch of them together, we can get some very interesting data on how rufuss is actually being used. And then once we know how Rufuss is being used, and not just what Amazon's seeing of how Rufuss is being used. 'cause there's a big difference of what Amazon says and sometimes the reality of it.

Chris: Right.

Matt: Then we take that data and then we can act on it.

Chris: Yeah. And you're going to these different events, giving talks, talking to other people. Do you think there's going to be a glut of commentary on AI that'll turn off [00:09:00] people that aren't living and breathing it every day? Or do you think it's just gonna get to the point where it's such a necessity that people want the glut of information so that they can.

Keep talking about it so that they can just, by sheer force, it will pound the information into their heads so that they're equipped to make those decisions.

Matt: Yeah, tons of conferences. Obviously everyone's talking about ai. One of the ones I am super pumped for is, I mean we know Kevin always puts on a great event.

Chris: Mm-hmm.

Matt: He's actually shifted his whole focus on e-comm to basically ai. So he has E-Comm Mastery ai, as his main event this year.

Chris: Mm-hmm.

Matt: Super pumped for that one. We're gonna be there, we have a booth as well too, and there's over 40 people and I've seen and know a lot of these people too.

Like I've got to see their talks or some of their workshops that they've done at another event called Market Masters. And it's next level stuff like it is mind blowing, the stuff that they're sharing, some of the things we had to like swear to secrecy because it's that, that interesting and I'm [00:10:00] sure you'll get a piece of that, when you're attending e-Comm Mastery ai.

So, I'm super pumped for those.

Chris: For Nashville. Yeah. Trying to figure out if I can make that, it's kind of a tough time because it's a big window, after the Prosper Show, which is what's coming up next.

Matt: And Prosper show of course, as well too. Yeah.

That's a staple. Like we'll be there, we'll be sponsoring it as well. Looking forward to that one.

Chris: Yeah, I know it's all the rage. I just want to make sure, like if brands ask me questions about it, I can put pe people in different categories for of need.

Like you're strong on this ai search concept, but you're weak on this one, and so this is the right place to go. My job is to direct people when they ask.

Matt: Yeah, like looking at the lineup, like for AI specific, like if someone has, they want the latest and greatest ai, I would run straight to, uh, to Kevin's event.

Chris: Yeah. And now what other initiatives or programs are you guys doing at Product Pinion? That are new [00:11:00] 2026, whether it's AI related or not, versus last year, the last time we talked.

Matt: Yeah. A lot of fun stuff. So, something we've been working with is direct and great integrations. So just working with a lot of the apps that can generate these creatives so that it's a one click, Hey, I generated these creatives one click and now we're testing them with real people.

Right? So as easy as it is to make creatives. We sometimes lose sight that at the end of the day it is still, now people looking at the creatives and you wanna make sure what you create actually resonates with real human beings.

Chris: Yeah.

Matt: Wow. Right? Like what a concept. So we wanna make it like super easy that you can take what AI creates and see.

'cause a lot of times when you're in the thick of it and you're doing these AI creations, it's hard to take a step back and get a like a third party kind of look or a new pair of eyes. And being able to test things and seeing what shoppers actually resonate or don't resonate with can make all the difference on getting getting those conversions.

So integrations is a big one. We're [00:12:00] also opening up a lot of testing opportunities with other channels. So for example, we just recently launched, we soft launched Walmart. We haven't actually announced it yet, but it is actually live in everyone's account. So you can actually make Walmart search simulations and see how you can increase your click through rate there too.

And something that we are also working on is an interesting setup with, testing TikTok. And trying to figure out which influencers would actually resonate best with your audience before actually going out there and paying for these influencer videos and understanding which ones would actually make the most, uh, out of 'em.

So yeah. Yeah, a lot of, a lot of fun stuff coming up there.

Chris: That sounds valuable because there's so many influencers. How do you wait, which one? Has good track record with your kind of product or the best approach to marketing.

Matt: in this case, what we'll also be able to test is actually AI influencers.

So you don't even have a real person behind the scenes. It's literally like, Hey, will this AI influencer be better than this one? And figure out with statistical [00:13:00] significance, if so. So that's gonna be a lot of

Chris: fun.

Yeah, I keep, my phone keeps giving me AI realistic girlfriend suggestions.

I was like, wow.

That's interesting. So you call, like I'm thinking how do you marry old world with New World? Like you call your AI girlfriend and have a conversation about where you're going tonight, but you don't actually go anywhere or you go somewhere virtual. Like what?

Matt: Having options is not a bad thing.

Chris: Oh my god. So funny. Yeah, no, that's great. So you're testing search. And different techniques on Amazon, and then rolling out to the other platforms that you're integrating with because you have some data and some use experience on Amazon first, which in theory, that'll translate over to Walmart, Etsy, whatever I would assume.

Matt: Totally. Yeah. Yeah.

Chris: Well, Matt, you're an easy guy to find online and in person. Let's just say people don't [00:14:00] roll into Nashville or Vegas anytime soon. Where can they reach you?

Matt: Yeah, easiest is matt@productopinion.com and obviously we have a free trial there. If you haven't tried us out. You can talk to, uh, get, get 20 shopper opinions for free just by signing up.

So, hit us up there and always happy to help and talk, shop and share ideas.

Chris: Yep. And if you have cool stuff in your booth at Prosper, I'm gonna be hanging out there too,

Awesome. All right, thanks Matt Kostan of ProductPinion. Thanks again for listening in to Seller Performance Solutions. We'll be coming back next week with another great episode. Have a good one.

Matt: Thanks Chris. Cheers.