Generative engine optimization (GEO) platform Lorelight, is shutting it down – not because it failed, but because the problem it solved didn’t need solving, according to its founder Benjamin Houy.
“Customers were churning because the product didn’t change what they needed to do. They would pursue the same brand-building fundamentals whether they had the data or not,” Houy wrote in a blog post.
The big idea. Launched in April, Lorelight pitched itself as a “proactive AI brand monitoring” tool. Lorelight promised real-time alerts when large language models, such as ChatGPT or Claude, misrepresented a brand.
The goal: To help marketers control their brand narrative in the age of AI by detecting inaccuracies, biases, or outdated info in AI-generated responses.
Lorelight claimed to offer visibility into how AI models “interpreted” brands and give companies a chance to correct or influence that narrative before misinformation spread.
Why it failed. Lorelight could show where brands appeared (or didn’t) in AI answers, but that data rarely led to new action, according to Houy. After months of analysis, Houy found that the brands showing up most often in AI-generated results shared familiar traits:
High-quality, helpful content.
Mentions in authoritative publications.
Strong reputations and subject-matter expertise.
Houy wrote:
“It’s the exact same stuff that’s always worked for SEO, PR, and brand building.
“There was no secret formula. No hidden hack. No special optimization technique that only applied to AI.
“There’s no secret GEO strategy. AI models reward the same fundamentals that already drive SEO and PR.”
The bigger picture. Houy concluded that GEO makes more sense as a feature within existing SEO platforms, not as a standalone category. Building a dedicated tool for tracking brand visibility in AI responses simply didn’t deliver enough unique value to sustain a business, he said.
Established SEO platforms, including Semrush, have already begun expanding into AI visibility and brand monitoring, integrating features that help marketers understand how brands appear in generative search results.
What they’re saying. Many SEO practitioners applauded the candor, via comments on Houy’s LinkedIn post. Some of the reactions:
Lily Ray said the post was something “the industry needs to hear.”
Gaetano DiNardi called it “saying the quiet part out loud.”
Kristine Strange praised Houy’s courage to step away from the idea he believed in.
Randall Choh countered that LLM visibility is already driving conversions, citing data showing that ChatGPT-sourced signups convert six times better than Google traffic.
Panos Kondylis argued the GEO space is “premature” – visibility tracking is early-stage and most tools echo what SEO platforms already do.
Yes, but. Beware of confirmation bias. One tool’s failure (that you probably hadn’t even heard about before it shut down) doesn’t prove an entire discipline is worthless. It’s still early.
If you believe in the Gartner Hype Cycle, GEO may simply be passing through the Trough of Disillusionment – when inflated expectations crash and weaker players fold before the survivors evolve into something more durable.
Lorelight lived for about seven months – from its April launch to its October shutdown. Its quick demise may be more about timing than the longer-term viability of GEO.
Some advertisers are noticing oddly cropped product images in Google Shopping ads — and it turns out Google Merchant Center’s “Smart Cropping” feature is behind it.
Why we care. Smart Cropping, enabled by default, uses automation to zoom in on what Google determines is the most relevant part of a product image. While the goal is to improve ad visuals, the result can sometimes be awkwardly cropped images that don’t match the uploaded product photos.
The backstory. An email from Google explains that there’s no option in the Merchant Center UI to disable Smart Cropping. Advertisers must instead contact Google support to have it manually turned off for their account.
The tip-off. Zato Founder Kirk Williams first raised the issue after spotting unusual ad visuals despite correctly formatted image uploads. He shared the finding on LinkedIn — and Google’s response — with the PPC community.
The bottom line. If your Shopping ads look off, Smart Cropping could be the culprit. Check your visuals and reach out to Google support if you want the feature disabled.
https://i0.wp.com/dubadosolutions.com/wp-content/uploads/2025/11/1761939627888-1-rUHRhI.jpg?fit=720%2C290&ssl=1290720http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png2025-11-04 18:13:262025-11-04 18:13:26Google’s “Smart Cropping” may be trimming your Shopping Ad images
Google’s AI Overviews and AI-driven search are reshaping content creation, SEO, and user behavior.
As we watch this fascinating evolution of search – and continue to debate what we call this new marketing discipline (HubSpot is opting for AEO, or answer engine optimization) – I interviewed Aja Frost, senior director of global growth and paid media at HubSpot. Some of the topics covered in our interview:
The need to redefine success metrics for AEO, prioritizing visibility and share of voice
HubSpot’s experimental journey, including creating hyperspecific, data-rich content and optimizing for LLMs.
Traffic directly from LLMs converts about 3x better than traditional search traffic for HubSpot.
This transcript has been edited for length and clarity.
Danny Goodwin: Hey everybody, this is Danny Goodwin, editorial director of Search Engine Land, and, today I’m being joined by Aja Frost. We have an interesting discussion coming up about GEO, AEO, AI, and all the good hot topics. It’s great to meet you Aja. ’cause I’ve actually never, uh, run into you on the conferences or anywhere. So it’s really nice to connect with you.
Aja Frost: You know, Danny, I was gonna say, it’s nice to see you, which is my go-to if I’m not sure whether I’ve seen someone, I met someone before. I figured we had met because we definitely run in the same circles. But I’m delighted to be finally, officially making your acquaintance.
Danny Goodwin: Absolutely. Before we dive in for the people watching or listening, do you want to introduce yourself? Tell us a little bit about who you are and what you do?
Aja Frost: Yep. I am Senior Director of Global Growth and Paid Media at HubSpot. Global Growth is our catch-all for top-of-funnel non-paid demand, which largely translates to SEO and now AEO. And I’ve been at HubSpot for a little over nine years, which is about eight years longer than I thought I would be. For those who don’t know, HubSpot is the customer platform that powers 268,000 teams. And it changes, I would say, as a company, every few years, which is what has kept me there. I think we have had a really interesting journey to this point, and we are embarking on what I believe is the most interesting era of SEO, AEO, and really marketing yet.
Danny Goodwin: Absolutely. So, yeah, it is a very fun time and you’ve been around for a few years at this point, so very curious to get your take. So, we had SMX Advanced a while back, our conference returned in person and at that point in time I’m like, oh, this whole a AEO versus GEO versus whatever we’re gonna call a debate – it’s gotta be settled by the time like October, November comes around. And I’m surprised that it has not still been settled. So I’m curious from your perspective, where do you stand on that whole name debate? What are you calling it, you know, this new form of SEO, or if it’s some, even if you consider it a new form of SEO, you know, has been GEO, AEO, some people call it AI SEO. What are you kind of calling this practice right now internally and, and why have you settled on whatever term that is?
Aja Frost: Yeah, great question because this was the topic of much debate internally at HubSpot. I think we debated all of the names that you just mentioned and probably 10 more. And we ultimately landed on AEO, or answer engine optimization, because we think it best reflects how people are using AI and what businesses/brands should be doing in response. So I think SEO, you wanted to rank in the results, like that was pretty clear. Now you wanna be a part of the answer. And so answer engine optimization is the tactics, the plays that you run to show up as part of that answer. Also, it just sounds cooler than GEO in my opinion, but we’ll see how long the debate rages on. I have learned not to underestimate how long people in our particular world can spend haggling and debating this type of thing.
Danny Goodwin: Yes, I know it’s, it’s sort of like subdomains versus subfolders. If you’ve been around long enough, you’ll know what that means and how long that debate has been going on. And I can’t even tell you, uh, more than a decade, I’m safe in assuming. Whatever we call it ultimately or whatever it gets decided it is called, this does feel like a big transition point for search from traditional ranking search to AI search is more about retrieval. So for you, how has it changed the way you’re thinking about visibility and strategy?
Aja Frost: Yeah, we are very much thinking about AEO as an evolution of SEO, which I did my homework and I’m just a Danny Goodwin fan, so I know that I think we’re on the same page there. And yes, that was an intentional pun. I think one thing that has actually always been a very HubSpot philosophy is do what’s best for the customer. And that’s always overlapped really neatly with our SEO strategy. It’s also what Google has preached for many years – do what’s best for the customer. You may miss out on some short-term wins, but in the long run, your site is going to perform better. And that is at the heart of our AEO strategy. I also think that the three buckets of plays that we’re running are familiar from SEO. So the what hasn’t changed, but the how has, and I’ll go a click deeper there. Those three buckets for us are content, technical, and offsite.
Our content for AEO looks fairly different than it does for SEO. It’s much more specific. It’s much nicher and deeper. It’s structured differently. It’s written differently. But it’s always intended to be what’s best for the customer or best for the reader.
The second bucket is technical. And again, I think that Google indexes/ingests content differently than AI bots do. And so we need to adjust our technical strategies to match while not doing anything that’s harmful for GoogleBot, because of course we still care about Google.
And then offsite, one thing that is probably the clearest from SEO to AEO is the emphasis on brand mentions rather than links. And so we’re really shifting our offsite strategy to be much more about positive mentions in the places that AI is training and citing versus getting backlinks on high domain authority websites.
Danny Goodwin: That is a big shift. I think still a lot of people aren’t ready for. So much of the stuff the tactics have been ingrained for – and I forget, how long have you been doing SEO roughly?
Aja Frost: I’ve been doing SEO for a little over a decade.
Danny Goodwin: So SEO is probably about near 30 years old at this point.
Aja Frost: Oh, Danny, we didn’t say we were gonna talk about my age on the podcast.
Danny Goodwin: Hey. But yeah. Um, sorry about that.
Aja Frost: No, they’re all good.
Danny Goodwin: So yeah, I mean, it’s just like, there’s this kind of, this whole playbook I think that a lot of people are attached to. And change is scary for a lot of people. Rethinking that stuff is important because nothing is static. And especially right now things are just kind of chaotic. The amount of changes we’re seeing, it’s crazy.
Aja Frost: Oh my God. Change is so scary.I think change is scary for us. We also had the pressure of not just figuring this out for our own internal strategy, but for figuring it out for our customers. The strategy that we are shipping right now, I have a very direct line to our VP of product for our marketing hub. I also spend a lot of time with the head of product for content hub. Those two products basically represent your website and content strategy and HubSpot. Everything that we’re doing. I’m telling them about the stuff that’s working, the stuff that’s not working, so they can turn that into product learnings as quickly as possible. I think it is terrifying and exhilarating and exciting all at once.
Danny Goodwin: Yeah. And with that change, I think there’s a lot of rethinking about how we define success, right? So AEO is not going to be the same success metrics that we had with SEO. So how are you actually thinking about that right now? It used to be like, how many links can I acquire? But what are you thinking about now? What’s important? Is it visibility in a AI answers, getting citations or mentions the actual conversions from the traffic, which again, is not as large as traffic from search, but – there is debate over whether it’s higher quality at this point, which maybe we’ll get into a little bit later. How are you sort of defining success with AEO?
Aja Frost: This was also a topic of much debate, and we actually published the results on our Loop Marketing page. We have a new scorecard for how companies should be thinking about marketing in the age of AI. And AEO, which fits into this loop marketing framework has a few new North Star metrics.
The first, and the one that I would argue is the most important, is visibility. And it’s visibility and not traffic, or not citations, because visibility is what’s going to ultimately inform whether someone converts. And they might not convert in that session. They’re probably not gonna convert directly from their interaction with the LLM. We know that LLMs just are really bad at navigational search. And so they’re probably opening up a new tab or maybe two days later, five days later, going to the website. But the, the visibility is what informs what we care about, which is the conversion. So that’s number one.
That takes, by the way, a lot of education with your exec leadership. And I am very lucky to work at a company, whose leadership is deeply embedded in all these conversations, and I think gets it. But if you are at a company where your CEO is not reading Search Engine Land, it’s definitely worth doing a deep dive to help them understand why visibility is the number one.
Second is share of voice. So what is your visibility like relative to your competitors? And I think that’s a really useful benchmark. I know that there was a lot of coverage back in mid-September when ChatGPT really turned down the dial on visibility for brands. And if you are just looking at visibility, you might think, oh, something’s going haywire with my strategy. If you look at share voice and share voice is constant or growing, you know that you’re doing the right thing, agnostic of some of the algorithmic changes.
Then we get to mentions, or sorry, mentions goes into visibility, then we get to citations. How many times is your website used as a source in answer engine responses? And I think this is really important. I think a lot of brands go after citations first. I’m putting it third on our list. I think it is important because if you get the citation, what we have found is your average ranking and the response and the sentiment of that description, they’re both better, which makes a ton of sense. If you control the source, you’re always gonna say the nicest things about yourself and put yourself first. If you overindex on citations, however, you’re gonna miss out on a wide swath of visibility that I think is pretty critical.
Danny Goodwin: You’ve done a lot of experimenting, which I want to get into in a minute, with optimizing for LLMs and AI-generated answers. What ways do you see SEO and AEO being similar? And then maybe where do you see them separating a little bit?
Aja Frost: Yeah, I think this goes back to what I was talking about – solving for the customer or doing what is good for the end user. I think that is shared for SEO and AEO. And one of the questions you probably get, ’cause I get it all the time, is, well, if I do this for AEO, will it be bad for SEO? And my answer is always no. If you are doing, if you were rolling out an AEO strategy that is good for the end user.
So an example of what would be bad for the end user would be burying secret instructions in content for an AI agent. A good thing would be creating really helpful specific content that’s going to answer a really niche query that someone is asking ChatGPT. And as long as you are solving for that end user, I think that you’ll benefit in both disciplines. You’ll, benefit in answer engines as well as Google.
And then I think the three higher level categories of plays are similar, but where I think things get very different are, again, the content is just, we’re going from these very broad, high level topics, these ultimate guides, which HubSpot – this is a, I don’t know, a dubious claim to fame. But when I started an SEO at HubSpot, then I was telling the blog team what keywords I thought we should target and, and recommending search friendly titles. And I really liked Ultimate Guide. I just thought it sounded nice. So every title I recommended was Ultimate Guide, this Ultimate Guide that. And then of course, a lot of websites started using Ultimate Guide, and now I’ll click through the SERPs and I see Ultimate Guide. I’m like, I think this is my fault.
So you’re going from the ultimate guide to, you know, this is the exact use case that this exact persona wants to accomplish, and here’s how to do it, and here’s some original data that we’ve gotten from customers just like you. And if you come from an answer engine, it’s gonna be tailored exactly to what we know about you. And so it’s a very different style of content and content journey.
Yeah. Yeah, yeah, for sure. ’cause I, I feel like, and I’ve, I had this conversation not publicly, but there were conversations after the whole bruhaha about all the traffic. HubSpot lost when that, that came out on, I don’t even remember what month that was this year, earlier probably in the spring. And just how much traffic they were losing. Everybody was losing their minds over it. And I was like, wow. You know, you kind of forget the influence that HubSpot had on content marketing as a whole. Your playbook that you guys came up with was used by so many other websites. Like there’s just, you know, repurposed for their specific topic or niche or whatever. But yeah, like HubSpot, that playbook was huge for a lot of years. Right. I think that’s, that was started like right before COVID around that time and then just sort of exploded., Is that the right timeframe?
Aja Frost: I think it depends on what you are talking about. If you’re talking about inbound, inbound I think is really at the heart of the web. At least for a lot of companies that were publishing educational content and inbound goes way, way back. I think we have always been very much a build and public company and, and we share our successes and our strategies along the way. Which is what we’re doing right now with Loop Marketing. I think that has led to a lot of companies saying, oh, this was really successful for HubSpot, I’m gonna adopt it as well, which is good. That’s what we wanted.
But I also think that when we started seeing declines from the emergence of AI Overviews and the changing nature of Google, that was a bit of a bellwether for what I think a lot of websites are now seeing. And so one response could have been, oh, we’re not gonna build in public anymore. We’re gonna be very cagey about what we’re doing and what’s working. So that doesn’t happen again. But that’s obviously not what we’re doing. We’re trying to be even more transparent and helpful. I really hope and believe that loop marketing, which is not a replacement of inbound, but meant to be, again, an extension of and, and a really helpful framework for companies can play that role.
Danny Goodwin: So just going back to that, that traffic drop. I was basically told it was about an 80% traffic drop and you kind of helped the company through that. And now in LLM world, HubSpot is the most cited CRM, is that correct?
Aja Frost: Or the most visible CRM
Danny Goodwin: Most visible. Okay. Gotcha. All right. And, and obviously this is, again, this is a fairly new technology. So, when you were starting to approach optimization on LLMs and AEO, how did you start that journey? Like, what were the first few things that you maybe either thought about or tried that did or did not work?
Aja Frost: Yeah. Well, the first thing I did that I would really recommend folks do if they don’t have an AEO function already stood up was I, um, pulled together some of the ICS on our team that were already doing a lot of experimentation and research in their own time. In my day-to-day, I am usually working with managers or directors. I’m not super close to the work. But I knew that I needed to be really close to this and really help guide it. And so I said, the three of us, we’re gonna meet once a day. We are going to launch one experiment per week if we can. I’m working with the dev team so that whatever we need to do, we can execute as quickly as possible. And so we took a very experimental mindset from the get go.
What we started out with was how do we scale good quality data-rich content? We had been thinking, and I think most people thought about content, maybe in a month you put out 30 pieces. If you’re a news publication, you could be putting out hundreds. But we’re thinking in multipliers of tens most teams. And I think we need to be thinking in multipliers of hundreds or thousands. And so with the team, I wanted to figure out how do we create that content? How do we start relatively small? So like batches of 10, generated with AI reviewed by a human, and then how do we scale that over time? That I think has been very successful.
We’re still experimenting with the types of content that get the most visibility in answer engines. And so that’s what a lot of experimentation revolves around. We also did a lot of what I think of as good clean AEO. Making sure that we were using all the available schema types across our website, making sure that things were really well structured and that we’re leading with the answer. And each section of the page is semantically complete and things are formatted in a Q and A format. You know, a lot of things that I think are now becoming like the standard AEO playbook.
Danny Goodwin: So you mentioned content types. I know there’s been a lot of noise about how some people are abusing top X lists – the top 10 best insert thing here. Is that the sort of stuff you’ve been playing around with? When you say content form, is there anything you can share about what you found that works maybe better?
Aja Frost: Yeah, so I’m not thinking so much about top X for Y, although I think that that still very much has a place in people’s content playbooks. But what we’re really experimenting with is – Danny, what’s the last thing you did research with ChatGPT to buy?
Danny Goodwin: Oh, to buy?
Aja Frost: Yeah.
Danny Goodwin: Uh, it’s, it’s probably researching to find a hotel for Christmas.
Aja Frost: Okay. Find a hotel for Christmas. So the context that ChapGPT is going to have when it recommends a hotel for you is probably about how much money you typically spend based on some demographic data it’s collected about you, if you’ve done any hotel research in the past, where you’re going, obviously how long you’re gonna stay. Hotels, we wanna provide the answers for all of those contextual clues. So if I were a hotel and I was trying to show up in answer engines, I would be creating content that spoke to your particular persona type and your particular use case. Now, I think the challenge is doing that without that content being duplicative or spammy. And to do that, this is what we spend a lot of time on. What are all the data sources that we can ingest to feed these systems essentially, so that all the content is unique, it’s grounded in what we know the persona needs, and it’s not repetitive from page to page.
Danny Goodwin: As, as you’ve gone through this process, were there any maybe big surprises like, oh my God, I didn’t think that would work. Or is there just like any kind of aha! moments, um, as you’ve been doing all this optimization for AI answers?
Aja Frost: The hardest part has been the measurement. I think that we are still very much as an industry, and I know this ’cause I talked to a lot of AEO vendors, figuring out how to correlate the actions that we are taking with specific visibility increases. And it’s highly dependent on the prompts you are tracking. I think that leaves the room for uncertainty and ambiguity because what if you’re tracking the wrong prompts? Or what if you’re tracking the right prompts, but not enough of them? It’s far less clear to say “I did X and Y happened” than it was with SEO. And even with SEO, you know, we couldn’t run A/B tests. We are always doing look backs. There’s so many variables at play.
I talked about education with execs around why visibility is the most important. I think the other really important piece of education, not just for executive leadership, but for, SEO/AEO teams is getting comfortable with less data and fewer direct lines between what we’re doing and the results. So that’s been, I don’t know if that’s been surprising ’cause I think I knew going in that that was going to be hard. But as we’ve progressed and we’ve done more and more teasing apart, the impact of individual experiments has gotten harder and harder.
Danny Goodwin: So I heard through on background of getting this interview set up that you sort of have a formula for getting ChatGPT to recommend a brand. So I want to hear all about that. What can you tell us about that?
Aja Frost: Well, I think that many of the best tactics that we are successfully using are ones that I’ve already mentioned. So we’ve spent a lot of time talking about hyper-specific persona-centric content. What we’ve talked about a little less is the off-site tactics that we’re using. And what we’ve done is identified ChatGPT and Google, because those are priority engines, we’ve identified their top training and citation sources. And then we have put together a concerted strategy to show up as positively and frequently as possible in those places. And two big areas for us have been YouTube and Reddit, which probably won’t surprise anyone as being very influential for answer engines. I can go a little bit more into some of the things we’ve done there, if that’s useful?
Danny Goodwin: Yeah, I think so. There’s been some research done around how heavily cited Reddit and YouTube and a few other sites are. So yeah, I’d be kinda curious to know, like from a strategic standpoint, maybe like how you guys are approaching Reddit and YouTube.
Aja Frost: Yeah. Very different strategies for each and one big learning for us, I wouldn’t say this is in the last year because we’ve been very active on both platforms for several years, but, um, treating every social media platform as its own beast and really getting to know the lay of the land and understanding the culture and the rules and the unspoken rules before we engage. I mean, that’s just a general best practice for any community or social media site.
But on YouTube, uh, we have a large slate of owned channels from Marketing Against the Grain and HubSpot Marketing, to how to HubSpot, science of scaling. It really runs the gamut. And we, the global growth or SEO AEO team works really closely with the teams creating those conthat content to weave in organic mentions of the products where they make sense and make sure that we are creating content on topics that we know answer engines and people care about. We also have a lot of creator partnerships with folks who speak to our relevant audience and somewhat similar playbook there. We want organic, relevant, contextual mentions of HubSpot.
Danny Goodwin: So that’s like influencer marketing, that sort of thing when you say creator?
Aja Frost: Yeah. I think you could call it influencer marketing. I mean, we, we sign, um, multi-month sometimes one-year contracts with creators and, and say, you know, we will pay you X, Y, Z and, in exchange you will create content on these wide topics. Well, we give them a lot of editorial freedom, but you know. You’ll mention HubSpot in X videos, that sort of thing.
And then on Reddit, it is a much more advocacy and community-centric approach. And I should have shouted out HubSpot Media on the YouTube front. They are a fantastic partner to my team. On the Reddit front, we work really closely with HubSpot community, another internal team. And in the last year we became the co-moderator of HubSpot’s subreddit. And we have spent most of our time making that subreddit as productive and engaging as possible because what we’ve seen, which is really interesting, is that the more activity that happens in our HubSpot, the more positive mentions of HubSpot there are across Reddit. Because basically you’re creating a team of advocates who are really excited about your brand, your product, and then they organically go out into conversations on our sales, our marketing, our CRM, and they say good things about HubSpot. So, very, very different strategies, but both focused on getting the right people to say nice things about HubSpot.
Danny Goodwin: I think we touched on this a little bit earlier. Google search versus traffic you get from AI engines, it’s very different. It’s not as large. We’ve actually reported, in the last couple months, three different stories basically saying that traffic that you get from LLMs is either worse or about on par with Google search in terms of converting. I’m curious what you’ve seen there. Do you see that to be the case or do you see quality traffic coming through?
Aja Frost: Yeah, the traffic that directly comes from LLMs converts at about three times better than traditional search for us. So we’re definitely seeing higher conversion rates. And I, I’ve read the SEL stories. I was looking at the one you most recently published, which was like 900 e-comm website over the course of a year. I shared that with my team last week. I was curious whether the difference in conversion rates had anything to do with the difference in the type of product and the buying journey. Like, I think by the time someone is coming to hubspot.com from an LLM, they’ve done a lot of research, at least that’s what our analysis suggests. And so they’re much readier to convert than someone who might in the old world have been coming to the blog to download an ebook on content marketing. It’s been another really fascinating area to watch the industry debate because I’ve also seen several different, uh, different stats.
Danny Goodwin: Right. Yeah. Again, it’s very early and these are not large scale studies, it’s just sort of anecdotal I guess we would say. But any data, I think is useful ’cause at least it gets people thinking about all of these things and it’s gonna always go back to, it depends. It may be different for ecomm versus B2B or whatever the case may be. I think there’s still a lot that’s going to change and where AI is now. I even today was seeing somebody saying we’re at peak AI already. Like really? Like it’s, it’s two years old. Like, come on.
Aja Frost: Yeah. I would disagree with that. Yeah. I think there are, to your point, some things that could be step function increases in conversion rates. Obviously instant checkout, that’s huge. I think that, yeah, I mean this was obviously over the course of a year and I do remember seeing in the study that conversion rates had increased over time, maybe as people got more comfortable or familiar with ChatGPT. But instant checkout’s huge. I don’t know what adoption for Atlas is going to be or for any of these ad browsers to be fair. But agent mode or agentic checkout would definitely improve conversion rates. So I think we’re at the very early innings of this.
Danny Goodwin: Where do you think AEO as a practice will be at maybe a year from now? Do you think it’ll be kind of its own thing? Do you think it’ll be part of SEO and is there anything that you were maybe kinda excited to see happen from ChatGPT or some of these other engines that could make these systems even better?
Aja Frost: I think a lot hinges on when Google makes AI Mode more of the primary search experience. I don’t believe that you are going to get an AI-powered answer for every search. My belief is for navigational queries, at the very least, you’re probably always gonna have something that feels like the traditional SERP and that it gets you from point A to point B very quickly. But I think for a lot, if not most other searches, you will probably be in some form of AI Mode and at that point, SEO and AEO become merged because there is no real traditional SERP to optimize for anymore.
Danny Goodwin: Yep. Exactly. That’s sort of been my problem with this whole naming debate. If you’re gonna call it AI SEO, what happens if that search engine goes away? There’s no more, there’s no more SE in SEO.
Aja Frost: Totally. Yeah. But yeah, and also that doesn’t exactly roll off the tongue. Like I don’t wanna stand up and and say I am an AI SEO.
Danny Goodwin: Right. Exactly. So if you could maybe give people one AEO type of experiment you think maybe they could run before the end of the year to kinda get a feel for it or just anything that you think might be helpful for them to kinda experiment with. Is there anything maybe you could suggest to people like, try this tactic or this strategy or whatever?
Aja Frost: I think if you want a real project, then I would try creating those hyper-specific, very persona-focused pages. I think if you’re looking for something that you could run with and get live by the end of the week, use one of the many query fan-out tools that are available online. Take a page that already exists on your website, plug like a, a likely reasonable query that would lead someone to that page into a query fan-out pool, and then assess whether your page answers or has content for all of the subqueries that that pool provides. And if it doesn’t add them and then see does your visibility for that head question increase.
Danny Goodwin: Awesome. Any final thoughts? Anything we didn’t talk about that you’d love to comment on or leave people with some parting words of wisdom?
Aja Frost: Yeah, I would, I would be remiss not to direct people to hubspot.com/loopmarketing. We have spent a lot of time on AEO. Of course, AEO is one of the tactics in this new growth framework for the AI era, but there’s a lot more that we believe businesses can and should be doing to not just survive but thrive. Check it out. I think there’s a lot there.
Danny Goodwin: Awesome. And just, just for anyone who’s listening and doesn’t know what is loop marketing like, can you give us just a quick overview of what that is? ’cause you mentioned a couple times.
Aja Frost: Yeah. Loop marketing is a growth framework for businesses. There are four phases: express, tailor, amplify, and evolve. Each of those four phases has a host of plays and tactics. But the general idea is that, as the web changes, as folks go from progressing through this ever narrowing funnel to getting an answer in an LLM, then going to your Instagram, then reading a review and, and really having like a much more messy, much less linear journey, we need a new framework for marketing. And so this framework is an ever-evolving, much more flexible dynamic framework.
Danny Goodwin: Right. So it’s sort of like that old bendy straw, the messy middle as Google put it, I think. Right?
Aja Frost: Yes. Yes. I will say messy middle came up many times in our conversations around the loop.
Danny Goodwin: Yeah. Awesome. Alright, well that is all the time I have for you for today. It was a great conversation. I really appreciate you taking the time to chat with us. Look forward to seeing more from you in the future and wishing you nothing but success heading forward.
Aja Frost: Thanks so much, Danny. This was really fun.
Danny Goodwin: All right. Thanks. Aja. Bye everybody.
https://i0.wp.com/dubadosolutions.com/wp-content/uploads/2025/11/twp4ozh58oy-PQnfoh.jpg?fit=1280%2C720&ssl=17201280http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png2025-11-04 17:22:402025-11-04 17:22:40Aja Frost on AI search, content strategy, and AEO success metrics
AI search isn’t killing SEO. It’s forcing it to evolve into a new, multi-platform discipline called search everywhere optimization, where social and user-generated content (UGC) are the new trust engines driving discoverability.
When I presented this concept at brightonSEO San Diego, what stood out wasn’t just the excitement around AI.
What stood out was the unexpected convergence of ideas across sessions. You might expect every talk to center on AI, yet a broader shift was quietly taking shape.
Across these discussions, one message echoed clearly: social and UGC now shape which brands audiences trust and engage with.
Below are four recurring themes from those talks, along with post-event insights from each speaker on how marketers can apply a search everywhere mindset.
1. Search is not a platform, it’s a behavior
Search no longer lives in one box – and users aren’t just Googling anymore. They’re discovering through:
Conversations.
Communities.
Creators.
While AI platforms are becoming part of that journey, much of it still happens where authentic discussions thrive: Reddit, TikTok, YouTube, LinkedIn, and Instagram, to name a few.
Search has never been more multi-platform, multi-touch, or multi-intent.
Marketers must now adapt to fragmented journeys that may start socially, evolve through AI, and end in branded discovery.
Garg, founder and CEO of Writesonic, said it well when he recently shared with me:
“Your website is no longer your main asset – your presence across the entire web is. Brands optimizing only for Google are missing 40% of their audience who’ve already moved to ‘search everywhere.’”
My presentation defined this concept as search everywhere optimization, emphasizing that success depends on SEO, social, PR, and brand teams working together to drive unified discoverability.
Other speakers echoed these points, even if they used different language.
Liddell defines this similarly as “search everywhere” – where social, brand, and search operate together to drive discoverability.
Hudgens said, “Social is evolving to become the new open web,” citing data showing traffic and engagement growth from social ecosystems.
Blyskal quantified the behavior: AI platforms cite Reddit and YouTube way more than any traditional websites. More proof that discovery has evolved beyond Google’s SERP.
In speaking with Blyskal, head of AI strategy and research at Profound, he noted:
“Search everywhere isn’t a trend anymore, it’s reality. Our data shows that consumers are asking ChatGPT, Claude, and Perplexity the same questions they used to ask Google, but the answers are being built from fundamentally different sources. UGC platforms like Reddit now drive more influence in AI recommendations than most corporate websites because they represent unfiltered human experience at scale.”
2. UGC and social content drive modern discovery
User-generated content and social discourse have become the connective tissue of search.
From product reviews to LinkedIn posts to Reddit threads, these conversations shape what AI and many humans believe to be authoritative.
Social platforms are now the front door to search intent, sparking curiosity and building interest that eventually leads users to branded and organic experiences.
Blyskal’s analysis of 40 million AI search results found Reddit to be the single most-cited domain across ChatGPT, Copilot, and Perplexity.
While some shifts have occurred recently, he confirmed on Oct. 21 that “Reddit is still the most cited website overall in AI and is still second in ChatGPT.”
Garg echoed this finding, noting that Reddit and other community-driven content dominate citations across industries – a clear signal for marketers to engage where real conversations happen.
Liddell’s award-winning BullyBillows case study demonstrated how social-first content can drive measurable SEO impact, including:
A 65% rise in brand searches.
A 195% increase in “brand + keyword” searches.
A 139% lift in revenue.
Reynolds likewise emphasized the value of social resonance, recommending that marketers invest in content that performs well on social platforms, even if it underperforms in organic search.
Seer Interactive’s own data backs this up: while social generates 89% less traffic than search, it produces 20% more leads.
Together, this data proves that social and UGC are not just amplification channels. They’re search inputs themselves, and a core component of search everywhere optimization.
In a follow-up conversation, Hudgens – founder and CEO of Siege Media – remarked:
“Search traffic to LinkedIn pages is up significantly, and I expect it to continue to grow, eventually coming close to Reddit and Quora in impact on B2B. Brands need to be considering how they show up and contribute on LinkedIn in order to best impact all search surfaces.”
Many are seeing this firsthand in their analytics – clicks are declining even when rankings remain steady.
The real goal now is preference: being chosen, not just seen.
Both humans and AI systems increasingly value authenticity and consensus over keyword precision and link quantity.
Today, search visibility depends as much on how others describe your brand as on the content you create yourself.
Liddell frames this shift through the lens of preference = authority + trust + relevance.
Reynolds highlights the rise of community platforms – LinkedIn, Reddit, Slack, and WhatsApp – urging SEOs to focus on spaces where people share content with personal endorsement, offering more genuine reach than traditional formats that dominate the SERP.
Hudgens describes the 2021–2026 content marketing evolution from “high DR (domain rating) links” to “high influence mentions,” signaling that social proof and reputation now act as the modern PageRank.
Garg quantifies it: AI now weighs third-party mentions three times higher than a brand’s own website.
In short, as search engines are learning to mirror people, they trust signals, not tactics. This is the preference component of search everywhere optimization.
Liddell, co-founder and Search Everywhere director at Deviation, summarized it nicely to me, sharing:
“Brands can’t win on rankings alone anymore; they win on trust. Modern discovery happens where people talk, not where algorithms dictate – and that means investing in authentic UGC and social visibility is as critical to search as backlinks once were.”
4. Search everywhere success starts with breaking down silos
In 2025, silos remain one of the biggest obstacles to growth.
Many of our clients experience this firsthand – and other industry experts agree that maximizing discoverability now depends on cross-functional collaboration.
Search teams can no longer operate in isolation. PR, brand, and social teams all feed the trust loop that AI, search engines, and users rely on.
Future success will depend on these groups meeting regularly, sharing ideas, and aligning on shared goals.
My presentation emphasized building cross-channel roadmaps with social, content, PR, and paid to ensure each team’s efforts reinforce each other.
Hudgens showed that the future of content marketing lies in blending PR, organic social, thought leadership, and SEO – creating compounding impact instead of treating them as separate channels.
Reynolds underscored the need for shared metrics, measuring impact not in rankings but in trust, reach, and conversion.
The new search equation runs on trust
While the speakers offered diverse perspectives, they all agreed on one central truth: search success is shifting from gaming algorithms to authentically earning audience trust.
Reddit posts, offsite reviews, social media, and third-party references now serve as critical trust signals – not because they link, but because they validate and build confidence in a brand.
This shift – evident across all four takeaways, from breaking down silos to valuing preference over ranking – underscores a broader reality: search isn’t something people do anymore.
It’s something they experience, everywhere.
The brands that will thrive in this new era won’t be those with the most backlinks or the sharpest keyword strategy, but those whose audiences genuinely connect with and vouch for them.
Over the past year, Google Ads has increasingly embraced automation, shifting the account manager’s role in both practice and strategy.
The granular control and transparency we once took for granted are rapidly disappearing.
As 2026 approaches, it’s time to face reality – five PPC tactics are falling out of favor in the new era of automation.
1. Relying on phrase match keywords
Once the go-to option for advertisers who weren’t ready for a broad match strategy but wanted to expand search volume, phrase match has recently fallen out of favor.
Google continues to redefine how match types work.
Because Smart Bidding and broad match rely on multiple intent signals, these signals now match user intent more accurately than phrase match did under the same strategy.
When targeting a specific query, exact match tends to provide stronger control, while phrase match often returns ads for irrelevant searches.
As a result, phrase match has become both too limited to scale an account and not precise enough to maintain the level of control advertisers need in a keyword match type.
2. Skipping standard shopping campaigns
Although Performance Max has been Google’s main focus for some time, advertisers continue to see strong results from testing standard shopping campaigns.
This became even more apparent after the ad rank update at the end of 2024, which removed Performance Max’s built-in priority over standard shopping.
Since then, standard shopping campaigns have outperformed Performance Max in many cases.
Standard shopping also provides greater channel control and a clearer attribution path, as conversions typically come from direct clicks within the Google Shopping network.
While Performance Max now offers campaign-level search terms, standard shopping has long provided both that data and impression share insights at the product-group level – valuable for benchmarking and understanding competitive performance.
If you’re concerned about brand safety, standard shopping is the safer choice. It helps keep your ads off irrelevant or inappropriate placements across the Display Network or YouTube.
3. Making GA4 your primary conversion action
Remember the days of Universal Analytics, when Google would always advise advertisers to use UA conversion tracking as the primary metric?
It seems the guidance has gone back and forth ever since.
Ideally, your main conversion metric in Google Ads should align with account conversions to deliver real-time data signals for Smart Bidding.
GA4’s tracking pixel doesn’t provide that freshness – imported GA4 events are delayed in processing.
Additionally, GA4 attributes conversions to the date the conversion occurred, whereas the native Google Ads tag attributes them to the date of the ad click.
Third-party tools such as Elevar or Analyzify often provide the most reliable setup for accurate conversion tracking.
If a third-party solution isn’t feasible, Google increasingly recommends the Google and YouTube app as an alternative.
It’s relatively easy to configure, but avoid syncing products or shipping settings during setup to prevent duplicate products or overwritten shipping details in Merchant Center.
GA4 should still be linked for audience building and secondary reporting, but it’s best not to use it as the primary conversion metric.
It simply doesn’t deliver the real-time data accuracy needed for optimal Smart Bidding performance.
Performance Max campaigns tend to favor branded queries, so it’s important to segment branded terms rather than allowing them to run within broader campaigns.
This matters most when aiming for incremental traffic growth, not just conversions you would have earned from branded searches anyway.
Performance Max prioritizes easy wins, bidding heavily on branded terms and often inflating campaign-level ROAS, making results appear stronger than they actually are.
Separating branded traffic into a dedicated brand search campaign provides more control over both budget allocation and bid strategy for those terms.
However, there are factors to consider before excluding branded terms from existing Performance Max campaigns.
Doing so can affect performance, and the right approach isn’t one-size-fits-all.
Review:
The campaign’s age.
History.
Contribution to overall performance.
The share of brand traffic it drives.
In large accounts, for instance, if a single PMax campaign is responsible for most conversions and spend, it may be unwise to exclude branded terms immediately.
Likewise, in accounts with limited budgets, keeping branded terms within the same campaign may still make sense.
5. Over-pinning responsive search ads
The pinning debate has been around for a while, but more advertisers are now leaning toward fewer responsive search ad (RSA) assets instead of over-pinning existing ones.
This helps maintain control over messaging while still giving Google enough flexibility to test which headline and description combinations perform best – without overwhelming the system with endless variations.
And yes, the question always comes up, “What about my ad strength?”
Realistically, ad strength should be treated as a guide for creative quality, not a direct measure of performance.
While it can highlight issues such as limited variety or missing keywords, it does not directly impact ad rank or quality score.
Ad strength is a diagnostic tool, not a KPI.
Chasing an “excellent” score by stuffing headlines and descriptions can easily result in weaker performance for the sake of a vanity metric.
Don’t fight the machine – feed it
As 2026 approaches, the most successful account managers will be those who adapt to the new landscape.
The goal isn’t to fight automation but to feed it the right data.
Focus on high-value inputs and let automation do the heavy lifting – the most profitable PPC practices are the ones that save time, not consume it.
https://i0.wp.com/dubadosolutions.com/wp-content/uploads/2025/11/Google-Ads-tactics-to-drop-UcdwUF.png?fit=1920%2C1080&ssl=110801920http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png2025-11-04 13:00:002025-11-04 13:00:005 Google Ads tactics to drop in 2026
Grappling with innovation and changing consumer attitudes is second nature to marketers, who have already lived through many technological shifts over the past two decades. But forecasting where things are going is especially hard when it comes to modern AI, which has such unusual, non-deterministic properties. You can’t just extrapolate from the state of AI today to understand where AI is going to be in five years (or one…); during this sort of a platform shift, you need to take a deeper first-principles look.
Some things won’t change. Consumers will always want products, services and experiences that resonate and meet their needs. Marketers will always want easier, faster and more effective ways to connect with consumers. But the technologies that mediate that relationship are primed to shift in the coming years in major, unprecedented ways — impacting how marketers do their work, and the customer experiences they’re able to deliver.
How the marketer experience will evolve: Less rote work, more creativity
The history of marketing is built around constant evolution. But the scale and complexity of the change triggered by the rise of modern AI may test even seasoned customer engagement teams. To thrive, marketers need to open themselves up to new skills, perspectives and capabilities that will allow them to do more with less.
This change is already underway. As marketers take advantage of AI, they’re spending less time on rote tasks (like manual message creation) and more on strategy and creative work — from brainstorming innovative campaigns to deepening their testing and optimization strategy. These efficiency gains will grow as AI becomes a more prominent part of the customer engagement process, allowing brands to set goals and guardrails, then empowering their AI solutions to independently consume context, make decisions, and act on marketers’ behalf.
Today, that might look like training basic agents on your brand’s voice to ensure that message content is consistently on brand. But as we gain trust in AI’s ability to operate unsupervised over longer time horizons and to handle complex projects, more marketers will be able to shift their focus to strategy and effective management of the AI resources at their disposal to enable AI decisioning and other essential optimizations.
How team experiences will evolve: Humans and AI agents working side by side
Marketing is a collaborative art, where building a successful customer engagement program often depends as much or more on marketers’ ability to work together effectively as it does on their individual skills. But while AI may help marketers to work with internal stakeholders more effectively, its biggest unlock is the ability to be a direct “teammate” to marketers themselves. And by leveraging AI’s ability to create countless agents that can support customer engagement, even entry-level marketers will likely find themselves essentially operating as a “manager” of a team of autonomous subordinates.
Imagine creating a whole team of agents, with one tasked with personalizing product recommendations, one that QAs messages to ensure they’re formatted and built correctly, one that handles translations and another that reports back at the first sign of campaign underperformance. By supplementing your existing capabilities with agents, you aren’t just reducing the burden on yourself and your human colleagues; you’re also building a digital institutional memory, training these “teammates” with context and goals and reward functions to be able to keep supporting your efforts and driving value even as human coworkers come and go and your team’s goals shift and evolve with time.
AI and customer engagement: How brands can win the future
For years, marketers have sought the ability to truly personalize communication on a 1:1 basis across an audience of millions, and to do it swiftly, efficiently and at scale. This was the Holy Grail of marketing, but due to the limitations of technology it simply wasn’t achievable for even the most advanced teams. That’s all being made possible by AI decisioning, a powerful new type of functionality that can force multiply brands’ marketing performance and creative impact while delivering what their customers want and need.
Previously, a brand trying to win back lapsing customers had a long journey ahead of it. It might start by leveraging a churn propensity model to identify which customers are most likely to churn, then use a product prediction model to figure out what products to highlight in order to tempt them to return. From there, they’d need to run a series of A/B tests in order to figure out which offers and channels will work best. But while taking that approach is a traditional best practice, it only got brands so far — they could target micro-segments on the right channel with the right offer, but truly 1:1 engagement was still out of reach.
AI decisioning represents a new way forward when it comes to personalization. This approach leverages reinforcement learning, where AI agents learn from consumer behavior and learn over time how to maximize rewards (such as conversions or purchases) in order to optimize the KPIs that have the biggest impact through ongoing, autonomous experimentation. That means AI decisioning can seamlessly determine not only the next best product offer for those lapsing users, but also the best channel, the optimal time of day or day of week, the frequency that makes the most sense, the message most likely to drive ideal outcomes, and any other dimension that could impact whether a recipient takes a given action.
Even better, because AI agents are constantly experimenting in the background, the model can continuously adapt to shifting consumer preferences and behavior. And because these models use first-party data about every available customer characteristic, AI decisioning makes it possible to engage with individuals in a true 1:1 way, rather than relying on segments. The result is exceptional relevance and responsive experiences for individual consumers, something that’s only possible because of AI.
Final thoughts
With any major technology shift, it isn’t enough to just plan for the obvious outcomes — you must ensure you can react effectively to the changes that no one knows are coming. To succeed, brands need to pay careful attention to the arc of this new technology. Responding to a platform shift can’t be a one-and-done thing, and brands that create a five-year plan without building in regular pulse points and adjustments are going to quickly find themselves falling behind their more agile, flexible peers.
To see the full benefit of AI in their customer engagement efforts, brands also need to look beyond AI. After all, AI isn’t a shortcut, it’s an amplifier — and the AI you use for customer engagement is only ever going to be as good as the infrastructure supporting it. An exceptional AI feature isn’t going to feel exceptional to consumers if it’s built on architecture that can’t take action in real time or can only deliver experiences in a single, prescribed way. Make sure your AI tools are built on a strong foundation and have the infrastructure they need to shine; otherwise, you may never fully achieve what’s possible.
Curious to learn more about how Braze is thinking about AI and customer engagement? Check out our BrazeAIᵀᴹ page.
https://i0.wp.com/dubadosolutions.com/wp-content/uploads/2025/11/Braze-20251104-selheader-v2-scaled-4WyNEg.jpg?fit=2048%2C1152&ssl=111522048http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png2025-11-04 12:00:002025-11-04 12:00:00Winning the platform shift by Braze
But no major AI platform has confirmed that they use it.
Not yet, anyway.
And there’s no evidence that any major large language model (LLM) actually uses it when crawling.
So, why are some SEOs and site owners already adding it to their sites?
Because LLM traffic is projected to explode over the next few years.
Which means AI models could soon become your biggest traffic source.
Remember: robots.txt was once optional, too.
Today, it’s essential for managing search crawlers.
LLMs.txt could follow a similar path — becoming the standard way to guide AI to your most important content.
In this guide, you’ll learn how llms.txt files work, the key pros and cons, and the exact steps to create one for your site.
You’ll also see different llms.txt examples from real sites.
First up: a quick explainer.
What Is LLMs.txt?
LLMs.txt is a plain-text file that tells AI models which pages to prioritize when crawling your site.
This proposed standard could make your content easier for AI systems to find, process, and cite.
Here’s how it works:
You create a text file called llms.txt
List your most important pages with brief descriptions of what each covers
Place it at your site’s root directory
In theory, LLM crawlers would then use the file to discover, prioritize, and better understand your key pages
For example, here’s what Yoast SEO’s llms.txt file looks like:
Does LLMs.txt Replace Robots.txt?
Short answer: No.
They serve different purposes.
Robots.txt tells crawlers what they’re allowed to access on a site.
It uses directives like “Allow” and “Disallow” to control crawling behavior.
LLMs.txt suggests which pages AI models should prioritize.
It doesn’t control access — it just provides a curated list. And makes it easier for crawlers to understand your content.
For example, you might use robots.txt to block crawlers from your admin dashboard and checkout pages.
Then, use llms.txt to point AI systems toward your help docs, product pages, and pricing guide.
Here’s a full breakdown of the differences:
LLMs.txt
Robots.txt
Purpose
Provides a curated list of key pages that AI models may use for information and sources
Sets rules for search engine crawlers on what to crawl and index
Target audience
LLMs like ChatGPT, Gemini, Claude, Perplexity
Traditional search engine bots (Googlebot, Bingbot, etc.)
Syntax
Markdown-based; human-readable
Plain text, specific directives
Enforcement
Proposed standard; adherence is not confirmed by major LLMs
Voluntary; considered standard practice and respected by major search engines
SEO/AI impact
May influence AI-generated summaries, citations, and content creation
Directly impacts search engine indexing and organic search rankings
Layout and Elements
So, what goes inside this file — and how should you structure it?
LLMs.txt should be created as a plain-text file and formatted with markdown.
Markdown uses simple symbols to structure content.
This includes:
# for a main heading, ## for section headings, ### for subheads
> to call out a short note or tip
– or * for bullet lists
[text](https://example.com/page) for a labeled link
Triple backticks (“`) to fence off code examples when you’re showing snippets in a doc or blog post
This makes the file easy for both humans and AI tools to read.
You can see the main elements in this llms.txt example:
# Title
> Description goes here (optional)
Additional details go here (optional)
## Section
- [Link title](https://link_url): Optional details
## Optional
- [Link title](https://link_url)
Now that you know how to format the file, let’s break down each part:
Title and optional description at the top: Add your site or company name, plus a brief description of what you do to give AI systems context
Sections with headers: Organize content by topic, like “Services,” “Case Studies,” or “Resources,” so crawlers can quickly identify what’s in the file
URLs with short descriptions: List key pages you want prioritized. Use clear, descriptive SEO-friendly URLs. And add a concise description after each link for context.
Optional sections: Consider adding lower-priority resources you want AI systems to be aware of but don’t need to emphasize — like “Our Team” or “Careers”
To put all the pieces together, let’s look at some examples.
Here’s how BX3 Interactive, a website development company, structures its llms.txt file:
It features:
The company’s name
Brief description
List of key service pages with URLs and one-sentence summaries
Top projects and case studies
Citation and linking guidelines
BX3 Interactive also includes target terms and specific CTAs for each URL.
If adopted, this approach could shape how LLMs reference the brand, guiding them toward BX3 Interactive’s preferred messaging and phrasing.
LLMs.txt files can also be more complex, depending on the site.
Like this example from the open-source platform Hugging Face:
It organizes hundreds of pages with nested headings to create a clear hierarchy.
But it goes well beyond URL lists and summaries.
It includes:
Step-by-step installation commands
Code examples for common tasks
Explanatory notes and references
This way, AI systems would get direct access to Hugging Face’s most valuable documentation without needing to crawl every page.
This could reduce the risk of key details getting missed or buried.
Keep in mind that the ideal structure depends on the scope of your site. And the depth of information you want AI to understand.
It’s possible that an llms.txt file could boost your AI SEO efforts over time.
But that would require widespread adoption.
No major AI platform has officially supported the use of llms.txt yet.
And Google has been especially clear — they don’t support it and aren’t planning to.
But big players like Hugging Face and Stripe already have llms.txt files on their sites.
Most notably, Anthropic, the company behind Claude, also has an llms.txt file on its website.
If one of the leading AI companies is using it themselves, it could mean they see potential for these files to play a bigger role in the future.
Note: While Anthropic has an llms.txt file on its site, it hasn’t publicly stated that its crawlers use or read these files.
Bottom line?
Treat llms.txt as a low-risk experiment, not a guaranteed way to boost AI visibility.
Potential Benefits
Right now, the benefits are theoretical.
But if llms.txt catches on, you could benefit in multiple ways:
Control what gets cited: Spotlight your blog posts, help docs, product pages, and policies so AI tools reference your best pages first instead of less important or outdated content
Make parsing easier: Your llms.txt file gives AI models clean markdown summaries instead of forcing them to parse through cluttered pages with navigation, ads, and JavaScript
Improve your AI performance: Guide AI models to your most valuable pages, potentially improving how often and accurately they cite your content in responses
Analyze your site faster: A flattened version of your site (a single, simplified file listing your key pages), makes it easier to run a keyword analysis and site audit without crawling every URL
Key Limitations and Challenges
The skepticism around llms.txt is valid.
Here are the biggest concerns:
No one’s officially using it yet: No major platforms have announced support for these files — not OpenAI, Google, Perplexity, or Anthropic
It’s a suggestion, not a rule: LLMs don’t have to “obey” your file, and you can’t block access to any pages. Need access control? Stick with robots.txt.
Easy to game: A separate markdown file creates an opportunity for spam. For example, site owners could overload it with keywords, content, and links that don’t align with their actual pages. Basically, keyword stuffing for the AI era.
You’re showing competitors your hand: A detailed llms.txt file hands your competitors a lot of info they might have to use dedicated tools to get otherwise. Your site structure, content gaps, messaging, keywords, and more.
Creating an llms.txt file is pretty simple — even if you don’t have much technical experience.
One caveat: You may need a developer’s help to upload it.
Step 1: Pick Your High-Priority Pages
Start by selecting the pages you want AI systems to crawl first.
Pro tip: Don’t dump your whole sitemap into your llms.txt file. Focus on your most valuable pages — not an exhaustive inventory.
Think about the evergreen content that best represents what you do — your core product pages, high-value guides, FAQ sections, key policies, and pricing details.
For example, BX3 Interactive lists this web development service page first in its llms.txt file:
Why? Because it’s a core service they offer.
And by featuring it in llms.txt, they’re signaling to AI crawlers that this page is central to their business.
Step 2: Create Your File
Next, open any plain-text editor and create a new file called llms.txt.
Options include Notepad, TextEdit (on Mac), and Visual Studio Code.
Pro tip: Don’t just list bare URLs. Add a brief description for each one that explains what the page covers and who it’s for. This context could help AI understand when and how to cite your brand.
Not comfortable with markdown formatting?
Ask your developer to handle it (if you have one).
Or let an LLM do the work — ChatGPT and Claude can generate these files instantly.
Here’s a prompt to get you started:
Create an llms.txt file in markdown format using this information:
Company Name: [Your Company Name]
Company Description: [One sentence about what you do]
Important Notes (optional):
[Key differentiator or important detail]
[What you do or don’t do]
[Another key point]
Products/Services
URL: [https://yoursite.com/product-1]
Description: [What it does and who it’s for]
URL: [https://yoursite.com/product-2]
Description: [What it does and who it’s for]
Blog/Resources
URL: [https://yoursite.com/blog-post-1]
Description: [What readers will learn]
URL: [https://yoursite.com/blog-post-2]
Description: [What readers will learn]
Company Pages
About: [https://yoursite.com/about] – [Company background and mission]
Contact: [https://yoursite.com/contact] – [How to reach you]
If setup is quick and you’re curious to experiment, it’s worth doing.
Worst case, nothing changes.
Best case, you’re ahead of the curve if AI platforms start paying attention.
In the meantime, don’t neglect proven SEO fundamentals.
Structured data, high-authority backlinks, and helpful content are what help AI — and traditional search engines — understand, trust, and surface your pages.
Want to boost your AI visibility now?
Check out our AI search guide for a framework that’s already working.
http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png00http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png2025-11-04 01:59:392025-11-04 01:59:39Does Your Website Need an LLMs.txt File? + How to Create One
Google has added a new user agent to its help documentation named Google-CWS. This is the Chrome Web Store user agent that is a user-triggered fetchers.
More details. Google posted about the new user agent over here, it reads; “The Chrome Web Store fetcher requests URLs that developers provide in the metadata of their Chrome extensions and themes.”
What are user-triggered fetchers. A user-triggered fetchers are initiated by users to perform a fetching function within a Google product.
The example provided by Google was “Google Site Verifier acts on a user’s request, or a site hosted on Google Cloud (GCP) has a feature that allows the site’s users to retrieve an external RSS feed. Because the fetch was requested by a user, these fetchers generally ignore robots.txt rules. The general technical properties of Google’s crawlers also apply to the user-triggered fetchers.”
Why we care. If you see this user agent in your crawl logs, you now know where it is from. The Chrome Web Store fetcher requests URLs that developers provide in the metadata of their Chrome extensions and themes.
https://i0.wp.com/dubadosolutions.com/wp-content/uploads/2025/11/google-cws-agent-scaled-ErMzpH.webp?fit=2048%2C526&ssl=15262048http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png2025-11-03 16:02:452025-11-03 16:02:45Google adds Chrome Web Store user agent
Search is changing faster than ever – and 2026 may be the year it fully breaks from the past.
Over the last year, AI has reshaped how people discover, decide, and convert, collapsing the traditional customer journey and cutting touchpoints in half.
AI-powered assistants and large language models (LLMs) will handle roughly 25% of global search queries by 2026, per Gatner, replacing many traditional search interactions.
We’re already seeing the effects. Traffic from LLMs is climbing at a hockey-stick pace, signaling a massive shift in how users find information.
To stay competitive, marketers need to build strong content and experience flywheels, as answer engine optimization (AEO) and generative engine optimization (GEO) become critical priorities.
Bruce Clay, founder and president of Bruce Clay Inc., predicted:
“AI-powered search is expected to cause traffic to continue to drop for many sites, creating a disturbance in the force.”
Adopting AI isn’t optional – it’s foundational.
Yet most marketing systems weren’t designed to operate in an AI-first world.
Disconnected tools and data silos make orchestration difficult and create inconsistencies that derail performance.
To succeed in 2026, brands will need integrated, cross-functional, omnichannel systems that connect data, content, and customer experience.
Building a resilient digital presence for 2026
Preparing a brand’s digital presence for an AI-driven world means rethinking data, tools, and customer experiences while presenting a clear, consistent brand story.
The goal is to deliver personalized content and be ready for agentic experiences, where AI assistants act on behalf of users.
This shift begins with the evolution of search itself.
The biggest change is moving away from a simple query-and-response model to a more dynamic, reasoning-driven conversation.
Traditional search was like a game of chess – discrete and predictable. AI search, on the other hand, is more like a jazz concert – continuous and fluid.
The experience has shifted from browsing lists and visiting websites to receiving direct, synthesized answers.
Instead of matching keywords to an index, AI uses query fan-out, which involves:
Breaking queries into components.
Analyzing multiple sources.
Delivering a single, comprehensive answer based on consistent patterns.
With AI, the traditional marketing funnel is shrinking. AI search can move directly from intent to conversion in minutes, dramatically accelerating the process.
We’re already seeing three- to eight-times higher conversion rates from traffic originating in AI search.
According to Crystal Carter, head of AI search and SEO communications at Wix:
“Traffic from LLMs (like ChatGPT and Perplexity) is becoming increasingly distinct from Google search traffic, requiring separate optimization and analysis strategies.”
New types of intents, like “generative” (e.g., “create an image”) and “no intent” (e.g., “thanks”), now make up almost half of all LLM interactions and don’t require a website visit.
Search is becoming action-oriented.
As AI systems start booking tables, making appointments, and completing purchases, even transactional journeys may no longer end on your website.
Search ‘everywhere’ optimization: The new SEO
For brands, the goal is no longer to be a single destination. It’s to be present wherever your audience is.
That means becoming a trusted data source that powers the new, agentic ecosystem.
AI systems prioritize clarity, consistency, and patterns, so channel silos must give way to a well-integrated, omnichannel approach.
Ideally, AI agents should be able to access all your brand data and deliver complete, contextually accurate results based on user intent.
As Bill Hunt, president of Back Azimuth Consulting, explained:
“AI agents like ChatGPT will shift from answering questions to completing transactions. Both the Shopify connectors and feeds, as well as Walmart and Amazon saying they are Google killers. Being ‘callable’ through APIs and integrations will be as critical in 2026 as being crawlable was in 2010.”
In this new paradigm, websites are evolving from sales destinations to data and information repositories – built not just for human visitors, but for AI systems that retrieve, interpret, and act on that data.
7 key focus areas shaping marketing and search in 2026
To compete in 2026 and beyond, brands must optimize for visibility across every relevant platform.
Here are seven key priorities and emerging trends shaping the future of search and martech.
1. Strengthen technical SEO foundations for AI retrievability
The foundation of search is shifting from traditional crawlability to GEO.
The core principle of GEO is retrievability – ensuring that high-quality content is not only discoverable but also easily accessible and understood by AI models.
To prepare for this shift, your website should serve as a centralized data hub for your content and digital assets, enhancing the experience for both humans and AI systems.
Make sure to grant access to AI crawlers in your robots.txt file, use server-side rendering (SSR) for core content, and adopt progressive indexing protocols like IndexNow, used by Bing.
2. Build localized visibility in AI-driven environments
Local SEO has evolved – from data accuracy in its 1.0 phase, to profile completeness and engagement in 2.0, to personalized experiences in what’s now emerging as Local 3.0.
AI models, particularly Google’s AI Mode, increasingly cite local business information from sources like Google Maps and online directories.
That makes core local SEO practices – NAP consistency and Google Business Profile optimization – critical for maintaining AI visibility.
Pages with robust schema markup also tend to earn higher citation rates in AI Overviews, reinforcing the importance of structured data for local relevance.
The biggest challenge today isn’t just creating content – it’s creating a connected experience.
As companies integrate AI into their digital experience platforms (DXPs), the focus must shift from producing siloed assets to building a connected content flywheel.
That begins with a deep understanding of who your customers are and what they need, allowing you to fill content gaps in real time and stay present at every critical touchpoint.
DXPs are no longer static repositories. They’re evolving into intelligent, AI-native engines that proactively shape user experiences.
The ideal platform uses AI to create quality content at scale, powering a flywheel that delivers personalized, efficient, and well-governed customer journeys.
This is especially important for large brands and multilocation businesses, where updating hundreds of pages still requires manual, repetitive effort.
Here are the key steps to creating quality content and building a content flywheel.
Insights: Identify customer intent and content gaps
Your content strategy should be guided by real-time customer needs.
Use AI-powered tools to uncover the questions and challenges your audience is trying to solve.
Then analyze your existing content to identify gaps where your brand isn’t providing the right answers.
Creation: Develop deep, AI-structured content
To create content that performs well in AI search, start by assessing AI visibility and user sentiment.
Use AI to scale the development of deep, comprehensive content – always with a human in the loop.
Since AI engines draw from text, images, videos, and charts, your content must be equally diverse.
Just as important, it must be machine-readable so AI systems can synthesize and reason with it.
Prioritize an entity-based SEO strategy to build topical authority, and use comprehensive schema markup to help search engines understand your brand and content context.
Clearly structuring your data also prepares your site for advanced conversational search.
It ensures visibility in the next generation of AI-powered answer engines and readiness for NLWeb, the open protocol spearheaded by Microsoft to make websites conversational.
Establish a human-in-the-loop workflow to review, update, and refresh content regularly, keeping it accurate, relevant, and effective in answering user queries.
Publish from a centralized source to maintain consistency across owned channels, and adopt rapid indexing protocols like IndexNow to accelerate discovery and visibility.
Monitor and iterate
Continuously track visibility and performance within AI models by testing target prompts.
Deploy an agile strategy – as you distribute content, monitor results, experiment with new approaches, and refine continuously, the flywheel becomes self-sustaining.
Each cycle feeds fresh insights back into the system, helping your content strategy stay adaptable and build momentum over time.
“AI search engines synthesize across ecosystems, not just pages. Marketing leaders must ensure their digital footprint works as a unified system, not isolated campaigns,” Hunt said.
Businesses must maintain consistent, clear information across every channel.
Traditional SEO is giving way to relevance engineering – a discipline centered on systematically creating and structuring content for semantic relevance.
This approach helps brands navigate today’s increasingly complex query landscape.
4. Create a consistent, data-driven experience flywheel
While the content flywheel attracts visitors, the experience flywheel converts them – a critical function in an era of zero-click searches. It operates on a continuous feedback loop.
Strategy: Building an experience strategy starts with unified data from every customer touchpoint and channel. AI can segment this data to reveal audience expectations and friction points, helping shape a strategy grounded in real behavior.
Experience: AI can then put this data to work – connecting audience intent, personas, desired outcomes, and business goals to generate predictive insights that drive personalized and agentic experiences dynamically.
Conversion: AI also helps track the customer journey through the funnel across channels and touchpoints. Dynamic A/B testing and conversion rate optimization (CRO) can then be done at scale, tailored to audience segments and intent.
Iteration: The goal isn’t perfection but agility. Monitoring performance alone isn’t enough – iteration matters. Use data to make real-time pivots, refining your strategy with every new learning.
The experience flywheel becomes a self-reinforcing engine that continuously drives engagement, builds loyalty, and accelerates growth.
5. Use AI agents to orchestrate journeys and workflows
As AI-driven search becomes increasingly agentic, it establishes a new standard for the seamless digital experiences customers expect.
To meet this demand, brands must use journey orchestration and workflow automation powered by AI agents that guide users through connected, intuitive experiences.
The key is to deploy specialized vertical AI agents trained on your business data.
By orchestrating these agents across the customer journey, you can deliver hyper-personalized, omnichannel experiences.
This is only possible if your website and systems are ready to interact with AI agents.
For internal teams, AI agents also offer major opportunities to automate manual workflows across the entire marketing landscape.
6. Redefine KPIs for an AI-first performance model
As AI satisfies user intent more directly within search results, traditional metrics like rankings and traffic are losing relevance.
This shift means citation is the new rank, pushing teams to optimize content for retrievability rather than rankability.
As metrics like click-through rate decline in importance, new success indicators are emerging – including LLM visibility score, AI citation count, share of voice, and sentiment.
Success now depends on query diversity, or the ability to answer multiple related long-tail queries effectively.
“Traditional metrics like impressions, clicks, and click-through rates are becoming much more difficult to rely on as KPIs. They are still useful to look at, but marketers should renew their focus on human behavior. Share of Voice is one of the best KPIs to measure this new behavior. Companies that ignore visibility in AI-driven responses risk ‘feeding that territory’ to their competitors.”
7. Integrate systems and data to power a unified marketing infrastructure
A fragmented marketing tech stack with siloed tools creates inefficiencies and hidden costs.
Data fragmentation and manual processes increase operational expenses and derail integration efforts.
Shifting focus to an integrated marketing platform – and evaluating total cost of ownership – helps overcome these challenges.
An integrated solution provides the consistency, clarity, and unified data needed to keep your digital presence adaptive and competitive.
As we move into 2026, AI is not just another tool – it’s rebuilding the customer journey from the ground up.
With AI assistants expected to handle a quarter of all search queries, the traditional marketing funnel is shrinking.
The new landscape is defined by agentic, action-oriented interactions that can bypass websites entirely, demanding a fundamental strategic shift from every brand.
To stay visible and relevant, businesses must evolve from being destinations to being trusted data sources for AI.
That begins by fueling a content flywheel with deep, structured content accessible across every channel.
Once this flywheel attracts an audience, an experience flywheel – powered by unified customer data and an integrated, AI-native platform – takes over to drive conversion through deep personalization.
Ultimately, the brands that succeed will be those that embrace this new ecosystem.
They’ll replace outdated metrics, such as traffic, with new KPIs focused on AI visibility, tear down silos through integration, and prioritize delivering seamless, omnichannel experiences.
Thank you to Bill Hunt, Ray Grieselhuber, Bruce Clay, Crystal Carter, David Banahan, and Tushar Prabhu for their insights and contributions.
https://i0.wp.com/dubadosolutions.com/wp-content/uploads/2025/11/customer-journey-evolution-Ks4T8Y.jpg?fit=1850%2C550&ssl=15501850http://dubadosolutions.com/wp-content/uploads/2017/05/dubado-logo-1.png2025-11-03 15:00:002025-11-03 15:00:007 focus areas as AI transforms search and the customer journey in 2026
John Mueller from Google posted an SEO tip and reminder for those who use cloud services, such as AWS, Azure, Google Cloud or others, to host images, videos or other content. John explained that you should probably verify those within Google Search Console. This will give you the ability to track the performance of those files in Google Search, including any debugging information when necessary.
Of course, in order to do this, you need to be able to control the DNS and most give you the option to do that through DNS CNAME. So you can set up your DNS to control those files in that cloud environment. For examples, it can be images.domain.com or videos.domain.com and so on.
The advice. Here is John’s post on this on Bluesky:
If you’re using a cloud provider to host images / videos / other content, you can and should verify the host in Search Console, so that you’re aware of potential issues that affect Google’s crawling & indexing, & Safe Browsing. Use a DNS CNAME to the bucket, then verify with DNS.
Using your own hostname (something like content.your-site.com) means you can verify it in Google Search Console to get crawl errors and malware alerts. You can verify using DNS verification… or just verifying your main domain.
To do this, set up a CNAME entry for your domain name and point at your cloud provider’s bucket, eg “content.your-site.com” uses a CNAME for “your-bucket.clodstorage.com” (or “buckets.clodstorage.com”). Also, you will have to update all links in your site (ugh, I know).
You need to update all the links within your site so that users only find your content with your new hostname. For bigger sites, this is a hassle, I know. Search & replace, then double-check by crawling the main sections of your site (all templates, all important URLs).
Caveat: if you need to do this for images, and you care about Image search traffic, know that this will cause fluctuations in Google Images (images are often recrawled slower than web pages and need to be “re-processed” with the new URLs). It’ll settle down though.
Bonus: if you use something like “content.your-site.com”, you can just verify the main domain with DNS in Search Console, and get all data for your website + the content hosted there in a single property in Search Console.
AND THAT’S NOT ALL. IF YOU ORDER NOW, YOU ALSO RECEIVE the ability to migrate to another cloud storage provider without breaking a sweat. Map the CNAME to the new bucket (if the file URLs remain the same), use redirects (it’s your hostname). It’s not really your site unless it’s on your domain name.”
Why we care. It is super common these days for websites to use numerous cloud hosting services and products. So it is totally possible that you are missing out on data, analytics and useful debugging details within Google Search Console for those services.
Verifying them on Search Console should not be a big deal for your site’s administrator and should it should unlock a lot of useful information for you and your SEO team.
Your Competitors Are Already Optimizing for AI Search. Are You?
✓ Monitor how AI platforms rank you vs competitors in real-time
✓ Discover untapped AI visibility opportunities in your industry
✓ Track sentiment shifts across 5+ major AI platforms
director at Deviation, summarized it nicely to me, sharing:
