Hey, Scott Austin here.

I want to start today with a number from two of my clients. Right now, for these two clients, about 7% of their sales are coming from AI, not from Google, not from paid ads, but from people who has ChatGPT or Claude or Gemini a question and got pointed at these stores and bought something.

What that 7% does not include is shop. Shopify counts the shop app as an a generic channel in its reporting. I've strip that out. The 7% is only the outside AI platforms.

Now that's the high watermark across the stores I work on most or lower, but two stores 7% of revenue is not a rounding error, and it did not exist as a channel two years ago.

So today we're going to talk about answer engine optimization or aiteo what it is, whether it's worth your time right now and what the work actually looks like on a Shopify store.

I'm going to show you three real client stores. All three got the same onsite work for me. All three ended up in different places in AI results.

The Future of AI commerce

So that's the situation today. The stronger argument for paying attention to Aiteo is what the next few years may look like. Bain, the consulting firm, but out of forecast that I will play a role in most online shopping journeys by 2030. Most not a niche, not an edge case. The default path to buying something online is going to run through AI, and that is already underway.

Bain estimates that 30 to 45% of U.S. consumers are using AI tools right now to research products and compare options. Now, I just want to say that number again and put it in a little bit of context, because sometimes these numbers can be misleading. So what the stat is, is 30 to 45% of U.S. consumers. What the stat is not is 30 to 45% of shopping experiences, right?

So any given consumer is going to have multiple shopping experiences. So if you hear that 30 to 45%, you're like, oh, there's no way that 35 to 45% of shopping experiences start with AI. You're right. But 30 to 45% of users have started using it. Hopefully that makes sense. Now. Another stat Salesforce measured AI influencing $3 billion of US Black Friday sales in 2025.

So the question is not whether your customers will use AI to decide what to buy. A lot of them already are. The question is whether AI knows your brand. When the customer asks the AI engine.

Is AI Overhyped?

You will read headlines about AI referral traffic growing 138 percent year over year, or 393 percent in a quarter, or 8x on Shopify's own platform data. Those numbers are real. Adobe and Shopify both published them.

But they are growth rates, and growth rates off a small base are still low.

Here is the absolute number. AI referral traffic is still well under 1 percent of total retail traffic. So when somebody tells you AI has already transformed ecommerce, that is not what the data says. Not yet.

But here is where the data is interesting. The traffic that does arrive is unusually good. AI-referred visitors have been measured converting at 54 percent higher rates, and generating 53 percent more revenue per visit than non-AI visitors. Shopify's own Q1 2026 platform data showed AI sessions converting around 50 percent better than organic search.

Small channel. Better conversion.

AI is a small channel today. It is the fastest-growing one. It converts better than anything else you have. And independent forecasts say it is heading toward a meaningful share of ecommerce this decade.

That is not a reason to bet your business on it. It is a very good reason to plant a flag now.

How AEO is different from SEO

I want to draw the line between AEO and SEO, because they are related but they are not the same.

The old SEO model was links. A backlink was a vote. Somebody linking to you passed authority, and you went out and built links.

AEO has the same DNA: third-party validation, but the mechanics changed in a few specific ways.

1. The link is optional now. An AI reading a page does not need a link. If a trusted care guide says your product is good, that association gets absorbed whether or not anyone linked to you. A mention with no link was close to worthless in old SEO. In AEO, it is the main unit of value.
2. What gets said matters, not just that you were mentioned. A backlink carried no meaning. A citation does. If a review says a product is dust-free and safe for respiratory health, that is a specific claim the model can repeat back to a shopper. The words in the mention are part of the value.
3. Frequency beats any single hit. In SEO, you could sometimes rank on a handful of powerful links. In AEO, it is about how often, across how many trusted sources, your brand comes up alongside the topic. It is closer to reputation building than to link acquisition.
4. And the sources are not the ones you would guess. Studies of AI citations consistently find community sources at or near the top. Reddit, YouTube, and LinkedIn. Not Instagram, not TikTok. Those two get cited far less.

Reddit in particular has been the giant. Through the first part of this year, studies were finding Reddit accounting for a huge share of AI citations, in some measurements close to half of what Perplexity was pulling from.

And I want to spend a minute on Reddit, because two separate things are happening to it.

Here is why Reddit mattered in the first place. It was authentic. Real people giving real opinions with no incentive to sell you anything. That is exactly what an AI wants when someone asks which product is good, and it is why a Reddit recommendation carries weight that a brand's own website never will.

Then everybody figured that out.

The Verge reported on skincare brands, record labels, and weight loss apps seeding Reddit threads. An entire cottage industry of Reddit agencies appeared. One of them charges brands 2,500 dollars a month to produce somewhere between 35 and 100 brand mentions. Researchers at Cornell Tech found that as few as 13 words in a Reddit comment can steer an AI answer toward a particular product.

Think about what that does. The thing that made Reddit worth citing was that nobody was gaming it. That is getting less true every month.

Reddit is fighting it. In July, they announced they are now catching around 25,000 spam posts and comments a day, using their own AI to spot the coordinated patterns their older filters missed. Moderators are pushing back too. One community wrote that companies were using them for AEO, and restricted new posts to a single weekly thread.

Additionally, Reddit's citation share is wildly unstable. In the middle of August, its share of citations in ChatGPT search fell off a cliff, from around 3.8 percent down to about half a percent in the space of a few days. Forbes, Axios, and Search Engine Land all covered it.

Nobody knows why it happened. OpenAI has never explained it. The technical change everyone blamed occurred six days earlier and does not account for the sharp drop. The firm that measured it says it cannot rule out a problem with its own data collection.

And this has happened twice before. In September of 2025, Reddit's ChatGPT share fell from around 7 percent to about 1 percent, then recovered to around 3 percent within two months. In June of this year, it fell from about 7 percent to under 1 percent, then climbed back to around 5 percent. Both times it came back.

There is also evidence that ChatGPT never stopped reading Reddit. It kept pulling Reddit pages in heavily while citing almost none of them. One consultant found a single conversation where 84 of the 221 pages ChatGPT retrieved were Reddit threads, and not one of them showed up as a citation.

So I am not going to tell you Reddit is dead, because that is not what the data says. What the data says is that your visibility on Reddit can drop by 80 or 90 percent in a week, for reasons nobody outside OpenAI can explain, and it has now done that three times in eighteen months.

That is the lesson, and it applies well beyond Reddit. You do not control the platforms, you do not control the engines, and you will not get a warning. So if you were paying an agency 2,500 dollars a month to plant Reddit mentions, understand what you were buying. Those mentions are still sitting there. Whether they are worth anything this month is somebody else's decision.

One last piece of context. Google signed a deal with Reddit in early 2024, reported at 60 million dollars a year, for access to Reddit content. OpenAI signed a similar one, reported at 70 million. That is a big part of why Reddit got so dominant in the first place. And Reddit is reportedly weighing whether to renew with Google, because AI summaries have been cutting the traffic that used to flow back to them.

So the source mix you optimize for today is not the source mix you will be optimizing for in eighteen months. The best practices are going to change rapidly, just like everything with AI.

The AEO strategy for Shopify stores

Here is the framework I use, and it has two parts. 1 is on-site. 2 is off-site.

1. On-site controls how you are represented. Structured data, product content, FAQ content, buying guides. This is making sure that when an AI looks at you, it gets accurate, complete, machine-readable facts.

2. Off-site controls whether you are recommended. This is citations and third parties saying you are good.

Which matters more depends on the question being asked.

For factual questions, like what size does this come in, or is it waterproof, or what does it cost, structured data does the work. The AI just needs your facts.

For recommendation questions, like what is the best hiking boot for wide feet, or which espresso machine should a beginner buy, citations do the work. All the on-site structured data in the world will not make an AI call you the best. Only third parties saying so do that. 

Now, sequencing. Do on-site first.  It is controllable. You own it, you can do it in weeks; there are no gatekeepers, and very little budget is required.

Then, once on-site is clean, shift the bulk of your ongoing effort to off-site, because that is the slower, higher-ceiling work.

I covered the onsite work in Episode 182, a practical guide to Shopify's agentic storefronts. If the on-site side is new to you, go back and listen to that one first, then come back to this. Getting your product categories and your category metafields in place is the prerequisite for everything else in this episode.

In addition to structured data on product pages, other types of content in your store can be useful for AI, including: 

1. Buying guides. The how-to-choose X article. How to pick the right size. What to look for in your first one.
2. Comparison content. X versus Y. Your material versus the cheaper material everyone compares you to.
3. FAQ content, in real question-and-answer format, because so many discovery queries are literally questions.
4. Use case and problem-solution content, aimed at the problem the shopper has before they know the product category.

On the FAQ piece specifically, Episode 183 goes deeper. That one is about the Shopify Knowledge Base app, and the part worth your time is the query log, which shows you the questions AI is being asked about your store and, more usefully, the ones it could not answer. Those unanswered questions are your FAQ content, handed to you.

Three stores, same on-site work, different results

So let's look at how this a work performs in the real world. I did the same on site work across three of my client stores. Same approach to structured data, same product data, health standards. I use my app Data Fi. You hear me talk about it all the time on all three, which is what makes the comparison clean. Data fi is a product data management app. It helps you with the structured data that AI needs, and running it across all three stores means the onsite variable is controlled in a way it usually is not. When we're comparing brands,

I'm going to show you the structured data for all three on screen in a minute, and you will see it is good on all three of them. And yet the AI results are different. That difference is entirely off site. And that is the lesson I want you to take away from today's episode.

Store One: Mulberry Park Silks

So the first brand I want to talk about is Mulberry Park silk. And what we're going to do for each one of these three brands is pick a relevant AI query and show the results to you across ChatGPT Claude in Gemini.

Now, right now I am in a guest browser session and I'm not logged in. I think I actually had to make a fake account for Claude because I just couldn't do it anonymously. But what I'm trying to do is come in like a fresh new user with no background or history with any of these brands.

So for the query for Mulberry Park Silks, who sells silk bedding, I ask the question. Actually, it wasn't even a question. It's a statement I'm looking for high quality silk sheets and ChatGPT instantly recommended Mulberry Park silks. It saw this as a shopping query and recommended, Mulberry Park Quilt twice out of five recommendations. And you can see here it's cited Forbes as its list. And then it recommends it again talks about it a little bit.

So now that same query. I'm looking for high quality silk sheets in Claude and I once again get a list of products I see Mulberry Park Silks, the brand we're talking about. Thick silks, also quince lily silk. So there's there's four brands shown here. I think the other one showed, three brands. But you see, Mulberry Park silk shows up at the top. And this is, you know, a high quality silk sheets are like 5 to $800. So this is, you know, in the SEO world or the CPC world, this would be a very valuable click that you would be bidding on.

Now let's go into Gemini and here you can see Mulberry Park Silks was the number one. More Rybak silks was the number two. And then Lee silk was three. Silk. Silk was for mommy. Silk was five silk, silky silk, silky mommy silk. So you can see all the different brands in there.

So if we look at the product page on Mulberry Park Silks the 22 mom is the one that actually got most, recommendations inside of the AI engines. I have this tool in Chrome that lets me see the structured data on any page. So if I click on that, you'll see here that we've got lots of structured data on here. There's an FAQ structured data. There's a product a product group returns shipping reviews lots of structured data here. Right. And this is the work done by data Phi and other elements that we've added to the store to get it all to present this information in the structured way that I can read.

So if we look at the product, there's lots of information about the products here. You can see, you know, all these additional properties of, you know, the fabric details, colors, different highlights. There's lots of information for AI to pull from this structured data. So it knows a lot about these products. So we can have confidence in this product. So this is just to show you that Mulberry Park Silk has good structured data on their website.

That's not why they're ranking now. They wouldn't rank without this information that you need to be clear on that, right. You have to do this part. But what gets them, know, and I was really surprised the first time it did this. By the way, I actually did this with my client a couple weeks ago, which prompted me to make this episode because I learned so much in figuring out why did they rank so high.

Because you put in a generic query like, I'm looking for high quality silk sheets, and this brand shows up on all three engines. You're like, wow, they did something really well. And I had to go figure it out because I was like, don't think the structured data alone is going to do this right.

So what does do this is all the third party citations that Mulberry Park Silks gets. And I'm going to show you three of them here to start off with. first one is, from Forbes. So a you know, well known you know, publication national global. You know Forbes right. It's not you know, the Bridgeport news or whatever. So you know they listed them on this is is one of the best Best silk sheets on Amazon. Right.

They also got ranked by Women's Health on the eight best silk sheets of 2025. They're right there. And then they're also on U.S. news as the best silk sheets on this one. It's best overall. And this is just three examples, national media ranking them. And these aren't like influencers or YouTube people. This is like national media, which gets them really, really high citations so that the AI engines have a lot of confidence when it recommends them.

On this page here, this is their, press release page. And every time they get featured in something, they put it up here and you can see there's just lots and lots of them, Pages and pages of them. Because they're getting recommended all the time. Because they have a really good product. Right. You're not going to get recommended without a really good product, which is why I trust all these citations. Right?

So Mulberry Park Silks ranks really high inside of I for two reasons. One, they did the on site work. Two, they did the offsite work to get all those citations.

Store Two: ReptiChip

All right. Brand number two rep the chip. Actually, had JT, the owner of Rep, to chip on the podcast years ago. Now, we're actually going to bring him back pretty soon, I think, in the next couple months. But we're ready. Chip does is they sell coconut substrate for your reptiles tank. So turtles, snakes, frogs and that kind of thing.

Now, the query I did is I'm looking for substrate. So at the bottom of your tank. Right. Whatever material you put down there I'm looking for substrate for my ball Python A ball python is a type of snakes. I want to be a little bit specific now. You see here in chat GPC rep, the chip showed up number one right on a list of other products.

Also, we go over to Claude and rep. The chip shows up. Number three on on the list, but it's mentioned. And then it actually when I started doing this query, I started off with recommend some, coconut substrate. And I was like, that's too niche, right? Because there's other types of substrate. So I went up a level and made it more generic at the substrate, and they still ranked at the top, which I was very happy to see. And then Gemini put them at the top of the list.

So if we look at the structured data on a chip, just like we did on Mulberry Park, you're going to see they got, you know, lots of good structured data on their website. And there's lots of information about their products. It's all structured. The AI engines can find it. So you're doing a really good job with all that information.

But the on site structured data isn't enough. Right. And another thing they did on site that helped them is they also made for every single animal, that they could think of. Right. There's like 300 of them on their website now. They made a page for each and every one of them. So not only do they have structured data on the product page talking about the coconuts substrate, they also have a page for ball pythons, right? And for every other reptile you can think of. And on those pages include lots information about the ball python and then the recommended products, including the substrate for them.

So when someone asks of a specific question, I'm looking for a substrate for a ball python, I can go I'm looking for substrate for a ball Python. Does anybody mention ball pythons in their substrate? And here there's a page dedicated to that to help I get that answer.

But once again the onsite stuff is just the table stakes for rip the chip to be in the consideration set. And what they have is they have a set of external citations. Now this is substrate for reptile. So it's a much smaller world right. You're not going to get Forbes or, you know, Good Housekeeping to write about the best substrates for your reptiles. But there are publications out there that are reptile focused, and they do talk about and recommend reptile chip.

So here on Reptile Files you can see that reptile chip got recommended. As you know, the one of the top substrates. And then on small pet expert there's a, you know, best substrate for ball python and reptile chip shows up as the best value. And on pet educate. There's also an article about best substrates for ball pythons and reptile chip shows up.

So they have these third party citations. They're not national media brands but in their space, which is you know, let's just say it's less competitive than the silk sheet space. They've got enough citations to rank them to the top. So when people do the right queries inside of AI, their brand is getting recommended.

Store Three: Healing Home Foods

All right. Let's move on to our third brand. The brand is Healing Home Foods. They make, nutritious and good tasting snacks. And I pulled this line right off of their homepage. Great tasting snacks made using raw food techniques. And that's, you know, one of their important value props for the health of these products. And the gut health is the raw food techniques and the ingredients that are used to make them, but they also taste great. So I took that line right off their homepage, right.

And I went into ChatGPT and said, hey, I'm looking to buy great tasting snacks made using raw food techniques and healing home foods doesn't show up. I went in to Claude ask the same question. Healing Home Foods does not show up other brands to and I went into Gemini. Same thing right? No healing home foods. And this is the important lesson here You saw two brands Mulberry Park Silks and Rep two chips. Do all the great onsite work and had citations.

So when Healing Home Foods same situation on site. We've got lots of good structured data. You know, lots of information about our product and the ingredients and all the structured data is there that I could need to, you know, be able to understand everything about these products However, there are no third party citations for healing home foods. They're a small brand and haven't done a lot of PR work to get themselves included into the different websites that are ranking and rating these types of snacks, right?

So if we do a web search for healing home foods, you know, what you're going to see on here is results from the store, from stores they're selling in because they have they have a whole, you know, retailer relationship going on their social presences. But what you're not going to see is, you know, good housekeeping or healthy food.com. I don't know if that exists or not. It's made it up. You're not going to see any third parties ranking and rating. Actually there's one here. But I guess it wasn't strong enough to get them, ranked and I.

Healing Home Foods, which has a great product. Right. The customer reviews are absolutely fantastic, right. For them to succeed in the AI world, they would have to invest some time and energy into those third party citations.

The Citation Work

So all three stores are in the game, because all three have the on-site foundation. Where they place is being decided by forces that live outside the store.

That is the honest boundary of what any app, including mine, can do for you. The on-site half is the half you control, and it is the half worth doing first because it is doable. The other half is earned.

For a small brand with no PR team and no PR budget, like most Shopify stores, the top tier press is realistically out of reach. Those slots mostly go to brands with representation and existing relationships. So here is the tier that is actually reachable.

• Micro influencers and niche creators in your category. Many will post for free product. Their content gets indexed and uses the exact language you want associated with your brand.
• Registered dietitians, or whatever the equivalent expert is in your category, approached directly rather than through a magazine. You get the authority signal without the gatekeeper.
• Independent bloggers who publish their own roundups. Far more accessible than editors, and each one is a citation in exactly the right context.
• Reddit and community forums, done honestly. Show up as yourself, not as a fake customer.
• Small podcasts. Founders telling their story, which generates a mention, show notes, and often a transcript.
• Accurate listings on product databases and the newer AI ingredient scoring sites, where a clean label product can score well purely on merit.

This is a months-long grind, and it only works if the product is good enough that people want to talk about it.

How to tell if it is working

Do not guess at this. Measure it, and it costs nothing but time.

Pick five queries a real customer would ask that would lead them to your products. Mix the phrasings. For a reptile substrate brand, that might be:

• Is coconut substrate good for my ball python
• What is the best substrate for a ball python
• Best coconut substrate for reptiles
• What should I put in the bottom of my ball python enclosure
• A brand comparison query like ReptiChip versus coconut fiber.

Run all five in ChatGPT, Claude, and Gemini. Same queries, every engine. Then record:

• which brands get named
• in what order
• whether you appear
• what source the engine cites for the mention.

Score the results. Three points if you are named first, two if you are in the top three, one if you are mentioned at all, zero if you are absent. Add it up per engine, and you have a baseline number you can retest in ninety days.

Summary

I learned a lot in preparing for the podcast episode.  Here's my summary of AEO for Shopify stores today.

AEO is real, the audience is real, and it is small today. It is a growing channel; it converts better than most traffic, and independent forecasts put it at a meaningful share of ecommerce by 2030. Being early here is cheap. Being late will not be.

It is content-heavy and technical work, like SEO. But the differences matter. Mentions rather than links. What gets said about you, not just who points at you. Reddit and YouTube rather than Instagram and TikTok. And a discovery layer that rewards guides and comparisons rather than product pages.

Start on-site, because it is the half you control and you can have it done in weeks. Clean structured data, question-shaped content, buying guides that answer the category question rather than describing your product. That is the foundation, and it is what puts you in the game.  And if the structured data side is where you want help, that is what Datify does.

Then start the slow work of getting other people to say your name and recommend your products. 

Run the five query test this week and write down your score. Then you will know where you actually stand instead of guessing.

Thanks for listening.