Leveling Up SEO for the AI Era |Tom Critchlow
Hi, and welcome back to The Search Session! I’m Gianluca Fiorelli, and today I’m joined by Tom Critchlow to unpack the uncomfortable truths of modern SEO and how SEOs can step up from tactics to business impact. We’ll also tackle the future of AI answers.
Here’s what you’ll take away:
Power of Linking Out: AI search is stuck in a wall-of-text; linking out restores real user value.
Promise of Google Web Guide: as 10 blue links fade, LLM-led, structured, link-rich results point the way—though real-time delivery is still missing.
Uncomfortable Truth: real wins come from brand, product, PR—not classic SEO tactics. SEO diagnoses, while others fix to drive outcomes.
SEO’s Place: at scale, SEO lives in product, as multiple specialties—and AI will accelerate fragmentation.
Tactics Now: if you’re doing modern SEO, there’s nothing special yet for AI search despite fast ecosystem shifts.
SEO vs. GEO: neither is one thing; adapt by business type, with brand reputation shaping AI-surface visibility.
Tom’s vision: beyond query fan-out and Fast Search, Google will add precomputed LLM authority and reputation signals.
Key to reducing organic dependence: diversify and block AI bots to gain future bargaining power—even as Google dominates.
You’ll find more in the conversation - join us!
Video Chapters
Transcript
Gianluca Fiorelli: Hi, and welcome back to The Search Session. I'm Gianluca Fiorelli, and today we have a really, really exciting guest—someone I’ve been looking forward to having on for this session.
He’s currently the Executive Vice President of Audience Growth at Raptive, but he’s also an independent strategy consultant. He’s an alumnus of Distilled, an agency that doesn’t exist anymore but meant a lot in the SEO world.
He also worked for Moz—if you remember, back in the day—and he also had experience working at Google, and, if I’m not wrong, at DeepMind. So these are some interesting insights into his working past. And this person is Tom Critchlow. Hey. Hi Tom. How are you doing?
Tom Critchlow: Hi Gianluca, thanks for having me on. I'm doing very well. Just one small correction—I did work at Google, but not at DeepMind. Just to set the record straight.
Gianluca Fiorelli: Ah! I don't know why I had this sort of Mandela Effect in my memory.
Tom Critchlow: Back when I was working at Google, AI wasn’t really a big thing yet. But I did work briefly with the team that was developing quantum computers—so that was kind of fun.
Gianluca Fiorelli: Oh, that can be for the last part of our conversation—when we eventually touch on the future—because I have some not-so sci-fi ideas about how Google could use quantum computing to beat the competition even more.
How SEO Is “Complicated” Now
Gianluca Fiorelli: So let’s start. I’m going to ask you a classic question: How is SEO treating you lately?
Tom Critchlow: Um, I think we could maybe describe our relationship with Google these days as… complicated. I think we’ve gone from an era where things were somewhat predictable—where the status quo made sense. We kind of knew how things worked. There was a pretty healthy relationship for the web and for Google, for publishers, and for websites.
But if we rewind to 2022, 2023—when the Helpful Content Update started—ever since then, starting with the updates, the rise of Reddit, and now AI Overviews, etc., I think the relationship has become much more complicated. Much more fractured. Much more heated.
We’ve seen a lot of the dialogue between the search industry and Danny Sullivan, and some individuals at Google, become strained. I think, over the last few years, Google has become much more aggressive in rolling out some pretty extreme algorithm updates and UI changes across the board.
Gianluca Fiorelli: Yes, totally. I must admit, we also need to be honest and try to understand people like Danny Sullivan. We all knew him before he joined Google, and then we saw the difference once he entered Google—he was trying to be diplomatic. But I mean, we’re not diplomatic as an industry. We're all black or white: Either "Google is wonderful," or "Google is… not wonderful"—to avoid using stronger words.
But something just came to mind. As SEOs, yes, we have every reason to criticize Google—after all, around 80% of traffic still comes from it. But why do we put so much blame on Google, when many of the LLMs are even worse?
At least Google still gives us some data—we still have the Google Search Console. So even if the numbers are off or scaled weirdly, like with AI Overview traffic or AI Mode traffic—we still have traffic data. But with OpenAI, especially? Nothing.
Tom Critchlow: Yes, I think it’s the same reason nobody cares when Bing does an algorithm update, right? The traffic is just too small. ChatGPT is really the only AI platform worth paying attention to right now. Perplexity, Claude—they send almost no traffic. Usage is tiny compared to ChatGPT.
And I do agree—ChatGPT is actually worse than Google when it comes to linking out, which says something, considering Google’s track record with AI Overviews and AI Mode. I think we’ll eventually see more of a dialogue around this.
Why Links (and “Reasons to Click”) Will Come Back
Tom Critchlow: But also, I think this whole conversation about AI search is still very shortsighted. We are living through the biggest disruption in search that the industry has seen—at least in my opinion—since the birth of Google. And the pace of change isn’t slowing down.
There are no ads—yet. So if Google or ChatGPT wants to introduce ads into these new AI search experiences, they’re going to have to train users to actually click on links. You can’t have ads without links.
Right now, AI Mode is still this Frankenstein experience—it’s kind of the web, kind of an LLM. Everything’s going to change.
So I think I have a bit of a pragmatic view on it. Yes, the way AI Mode and AI Overviews handle linking out today is far from ideal. I also think the user experience is subpar. And I believe Google is going to come to the realization that links do have value.
Tom Critchlow: Actually, just this morning I was looking at OpenAI’s new brand ad campaign for ChatGPT. And the examples they’re using? I found and tested each one of them in ChatGPT myself. And honestly, the examples they’re using are literally better if you just ask for links from the web.
For example, one of the examples in this brand campaign shows a guy cooking a meal for a date night. He asks ChatGPT: "Give me a recipe that’s kind of cool, but not too cool—I want to play it cool with my date."
And ChatGPT replies, "Great! You should make this recipe," followed by five bullet points.
Stuff like: “Cook the pasta, cook the tomatoes, add cheese.” I mean, if I’m having a date night, I’m not cooking off of three bullet points! The response is literally better. I ran the same prompt myself and asked for a recipe from the web; the result was so much better—full step-by-step instructions, photos, and so on.
So it’s kind of a trite point, but I think it’s really important that these things are just better when they link out.
Right now, we’re living in this kind of uncanny valley where LLMs are so good at generating text that it’s become the default response type—they just spit walls of text back at you. And that’s okay for certain types of queries. But honestly, most of those walls of text responses would be so much better if they were enriched and augmented with relevant links. Like: “Here’s some links to go further.”, “Here are some links to dive into something.”, “Here are some links with photos, images, and a firsthand perspective.” All the stuff we know the web is actually good for. So I kind of feel like that’s going to come back around.
Gianluca Fiorelli: It still really bugs me. I wrote for our host, AWR, a post about Google Web Guide, which is still just in Search Labs right now. And I wonder why no one’s really talking about it. Because I actually think Google Web Guide is a good compromise between using Gemini for generating an introduction and for creating—thanks to its version of a query fan-out—a sort of big pillar page about the topic.
And then there’s another thing, which is great in the use of Gemini: creating the description of each search result. Instead of using the meta description or grabbing some text just because the keyword is inside, it creates something like: "In this page, this brand is talking about this and this." So it’s creating better information for the person.
Obviously, I think that could be a good use of search—one that is able to introduce quick answers while maintaining links, instead of creating something like AI Overviews, where the links are substantially invisible.
And then AI Mode—I mean, AI Mode could be something interesting, if it had the courage, let’s say, to go directly into what they want. Not just putting something out once in a while. Because I substantially think they already know what they want.
They showed us that already at Google I/O from a year ago. In the demo, they were already showing us what they thought was going to become AI Mode. So, go directly with that.
Tom Critchlow: I completely agree with you. There are two really important ideas here, and I’m glad you brought this up—because this is one of my strongly held beliefs: We are done with the age of “10 blue links.” That era—where you’d Google a query or a search term and get back a list of unopinionated, undifferentiated links—is over. Implicitly, what Google was saying with 10 blue links was: "Here are things that are relevant to your query." But it never explained why they were relevant.
For the last 25 years, Google has never told you why the link that is ranking number one for the query that you searched is relevant. So, implicitly, it's saying: “By showing up as number one, it's relevant, and you should click it.” But it never explained why, right?
And I think that using LLMs—they're very good at this. They're very good at saying, "Oh, you searched for this thing. These things are relevant, and here’s why." Relevant to why you asked the question: “You should click on this link.” Or, if you wanted a different kind of direction, “You should click on this one.” Or, if you wanted a video: “Go over here.” It can actually tell you why each result is ranking there.
So I completely agree, Gianluca. I think this is the future—we're never going to see the 10 blue links again. That era is going to die. But I do think links are going to survive—very similar to how they show up in Web Guide today, where they’re contextualized, and there’s a reason for linking and clicking to that link.
Now, the second most important point is: I think Web Guide is a really interesting thing from Google… but when you use it, it’s slow. It takes like three or four seconds to generate a response. And so again—back to this kind of uncanny valley—I think we’re living in an age where Google knows what the better user experience is, but it can’t deliver it fast enough.
With AI Overviews, the reason they use citations for links—rather than proper inline links—is because it’s faster. It’s faster to generate a response and then ground it with citations than it is to take all the webpages relevant to a query, synthesize them, summarize them, create that pillar page, and explain why each link exists. It’s just too slow, right?
Google—for 20 years—has prided itself on the speed of the response. And so, I think we’re, again, living in this kind of strange time where LLMs can be used to create a great search results page… but you just can’t do it fast enough to serve a traditional Google result. So, Google is somewhat stuck in this. That’s why Web Guide exists in its own tab.
That’s why AI Mode is its own tab. They’re getting in their own way. The question is: how do they get to the future, where they can provide a great search results page with links—and do it fast? I think it’s a technical challenge they’re facing right now. And I believe that’s going to get solved.
I mean, Google has the best engineers in the world working on exactly these kinds of problems. So I do have confidence and faith that as we move into 2026, we’re going to see more and more of that Web Guide style: "Here’s a link—and here’s why you should click it." I think that’s going to come back—into AI Overviews, into AI Mode; however, all of this evolves and blends together. I really believe that’s the future. But again—I’d encourage everyone to go play around with Web Guide. There are some really interesting ideas in there. But it’s also just too slow.
Gianluca Fiorelli: Yes, that’s why— for the people who haven’t tested it—when you open Web Guide, the first thing you see is the classic first results, the 10 blue links, as the filter web is thought to be. And then, when you’re starting to see these first classic search results, it loads the guide below.
And yes, I think that, at the end of the day, neither AI Mode nor Web Guide as they are will exist; everything will be merged, and we will see—whenever data and technology allow—something like an AI Overview for conversation inside the SERP, and the search results organized as if it were Web Guide.
Tom Critchlow: Yes.
Gianluca Fiorelli: But I think that, for people still thinking with nostalgia—which is cool until it becomes a sort of sickness— for the 10 blue links. We’ve been talking about zero-click search for years. In the zero-click search era, we were seeing first pages with only seven blue links, because everything else was a search feature.
And even with all these search features, people—maybe the smarter ones in the SEO industry—were already saying: “When you’re working in SEO, you don’t just rank with a search snippet; you have to start ranking with your video. You have to start ranking with your results inside People Also Ask, and all the other features.”So that was a good training for the situation we’re living in now.
What Really Moves SEO
Gianluca Fiorelli: But in this context of the introduction of AI search, where do you see the search industry among all the other—let’s use an improper word—channels for digital marketing?
What is the role? What is going to be the role of SEO now, when it’s so put into question because traffic is fading away? Visibility is a good metric, but it’s quite hard to sell it to your clients and your boss. Where do you see us?
Tom Critchlow: Yes, here is the uncomfortable truth: the best teams at producing SEO outcomes have not ever been SEO teams. I’m gonna say that again: the best teams at creating SEO outcomes have never been SEO teams.
The things that really move the needle for SEO are links, brand reputation, and authority. These are the fundamentals of SEO performance.
Sure—keyword usage and technical crawling—I’m not saying those things don’t matter. They’ve always been important, and you can get some wins there. But the things that really move the needle in terms of long-term SEO outcomes, SEO success, have always been the brand marketing team, right? The product team, right?
These are the teams that fundamentally create good products that people want to use; that create brands people want to care about and link to and talk about. The PR teams that go and get coverage in real news media outlets. That has always been true.
So the SEO industry has always had this problem where the SEO teams and practitioners understand Google better than everyone else. We understand keywords, we understand rankings—we understand all of this stuff way better than the other teams: the brand marketing team, the product team. But our ability to actually drive outcomes as SEO professionals has always been second fiddle to these teams that know how to pull the real levers that matter over long time periods.
And I think that’s true going into the AI world, too, right? We all fundamentally believe that brand reputation, brand coverage, product usage, and adoption—those things are going to continue to be what all the LLM models, however they measure them, really pick up as drivers of long-term success and performance.
You know—content chunking, writing self-serving LinkedIn articles, listicles, Reddit spamming—that stuff kind of works today, maybe a little bit. But if you’re a real brand, that isn’t where you’re putting your dollars. You’re putting your dollars into the brand team, the product team, the PR team, and the marketing team.
And so, I mean, we’ve had this crisis for a long time. I talked about this at the SEO MBA a lot. This has been my mission and quest for a long time: SEO professionals have got to level up.
You’ve gotta step up into executive roles, you’ve gotta step up into understanding the business ecosystem. You’ve gotta really understand things like brand marketing and product development. And if you don’t do those things, you’re going to get marginalized, right?
You think back—technical SEO used to be more important than it is today. That might be a controversial statement—people might not agree with me—but technical SEO used to be a big deal because websites had lots of problems, Google had problems crawling; it was a bigger piece of the pie. I’m not saying technical SEO isn’t important today, but it’s not as important as it used to be.
And if you look at the total amount of dollars that businesses invest, the proportion of that pie that’s going to technical SEO is shrinking. And so, again, if you think about the long-term outlook here—if you want to future-proof your career, if you want to get involved in what’s happening next—you’ve gotta think about the business ecosystem. You’ve gotta think about moving the levers that actually matter.
And if you are an SEO professional recommending to clients that brand reputation matters, then you’ve gotta take a long, hard look in the mirror and ask: “Am I, as a professional, capable of moving those needles? Am I capable of doing the things that really matter at the end of the day?” I think that’s an uncomfortable conversation, but it’s one we need to have as an industry.
Gianluca Fiorelli: No, I like uncomfortable conversations—because those are the ones that really make us advance—instead of the self-pity conversations: “We’re the worst or we’re the best.” Those are self-comforting, but maybe not the best for growth.
I’ll use myself—not as an example for all SEOs—but I think my evolution is relevant. I mean, it’s 20 years that I’ve been doing SEO, but obviously, let’s say, my evolution in the last six or seven years has been different.
And I think it’s interesting because for many, many years, SEO strategy was somehow a neglected area. And when I say SEO strategy, I mean giving insights to a brand from search about how to grow the visibility of the brand itself. That’s my idea of strategy around visibility.
But it has always been, let’s say, neglected—because SEO is very “Let’s do this, get results now,” etc., etc. Also, because this is what we have told brands we are about.
And I think this is maybe—at least for me—where I’m going. I’m going where I have the capability to understand how search works. I’m starting to understand—because in this field it is still very new—how AI search is working.
For instance, you were saying we can inform—obviously, the heavy lift is going to be done by the brand marketing team—but we can inform how to do that better, how to get better results, how to enhance the results thanks to our insights into how these machines are working.
For example, a very “stupid” area—coming back to the technical SEO part: maybe now, because there is this limitation of LLMs not rendering JavaScript (which will probably pass in the future)…
When we’re talking about brand SEO, rather than telling a brand marketing team what campaign to do, maybe some insights can be offered because we can understand trends. We can say: “We can help you make the brand even more unique and disambiguated thanks to knowledge graph, structured data, and these kinds of things.
A friend of mine was telling me, for instance, about a very famous Spanish brand of Inditex. So the guys of Zara—but it’s not Zara, which is in the Knowledge Graph for Spain, but it’s not in the Knowledge Graph for the US. Why? Because they didn’t care about creating good knowledge–graph–based, organization-structured data for the English language.
So this is because the brand’s name is a surname, and Google is not recognizing it as a brand. And so they are not appearing—you know, the Knowledge Graph is really strong for brand positioning on the SERP.
The New SEO Agency Model
Gianluca Fiorelli: And these are the things that we can help and support—more than saying, “You should do this,” or “We should do this digital PR campaign.” Yes, digital PR is… SEO? I don’t know if it is. It’s always been considered a part of SEO—but I think it’s PR, and PR is not really SEO.
Tom Critchlow: Yes, I think there are a lot of SEO agencies, in particular, that are having—well, this is a trend that’s been going on for 10 years now, Gianluca, right?
Let’s go all the way back. Let's rewind the clock right back to like 2009. SEO agencies used to be great at doing SEO because what you did was say: “Hey client, I’m gonna get you a bunch of article links and a bunch of directory links,” and that moved the needle. You could say, “I’m gonna get this page ranking higher.”
Then Panda came along, and the big Google algorithm updates of 2010–2011.
Gianluca Fiorelli: Also Penguin.
Tom Critchlow: And suddenly you’re in a world where SEO agencies are like, “Oh, you need content marketing—you need real, quality links.”
So Distilled and others, right—we launched into content marketing as the new thing: engaging infographics, link baits, all those things we used to do. That stuff would drive links, and links would drive SEO performance.
And that worked for a bit. But as that evolution happened—and as content marketing became more mainstream and harder—a lot of the smartest agencies realized: “Oh, this is just digital PR. This is real PR. There are teams that know how to do this—build relationships with journalists, get real coverage, real news stories, real links. And so it became that, right? Digital PR is really just PR.
The digital PR that a lot of SEO agencies were doing—and some are still doing today—was second fiddle to what real companies are doing.
And again, it goes back to my point: if you’re working as an SEO professional, your ability to actually move the needle on some of these metrics is somewhat limited.
So the best SEO agencies have realized this and done a couple of things. They’ve either built real PR capabilities, truly leveling up their digital PR and outreach efforts. Or they’ve pushed into a consultancy mindset—a bit like what you said, Gianluca—acting as senior advisors to brand campaigns, PR campaigns, and the rest. Or they’ve pushed into the technical side. Which—again—there’s plenty of dollars still left for technical SEO.
Certainly, you know, I think of Eli Schwartz with Product-led SEO—the whole idea that SEO is no longer a function that sits on its own. For scaled businesses, SEO is a function that lives inside the product organization.
And so, again, the smartest ones—both individuals and agencies—have realized that product-led SEO is a real thing. But it’s a real thing that lives inside the product team, not inside the SEO team or the marketing team.
And so again, you have this kind of divergence. We talk about SEO as being a single thing, but it’s many different things. And that fragmentation—those specializations in the industry—has been happening for a long time.
I think the era of AI search is only going to accelerate those trends. It’s going to push work toward product-led SEO, technical SEO, PR, and brand-led work. It’s really going to fragment, I think.
Gianluca Fiorelli: Yes, but it was fragmented—it’s clear. For instance, in the very beginning, I remember the person doing SEO was also doing paid advertising, and then that was the first big split: either you’re doing SEO or you’re doing SEM.
And then it started to be: you are specialized in technical SEO, you are specialized in content SEO—which somehow implies many other things that went very wrong. And then—for instance, myself—I define myself as a strategic SEO or international SEO, depending on the situation.
But somehow we also lived the other way around: SEOs trying to put their feet in every possible field of website work, putting SEO at the center of everything. So if someone was doing social media, SEO needed to inform social about what to do.
Content was never something related to SEO for me. Content marketing is a ship by itself—it cannot be “SEO in content marketing.” Also, because many times SEO ruined content marketing. And then all the things, like link building, became something that created agencies doing just that. And then evolving to digital PR, etc., etc.
I think this is somehow our own omnivorous tendency to appropriate everything—to be at the center of attention for everything in digital marketing.
SEO vs. GEO Is a False Binary
Gianluca Fiorelli: But anyway—you published, very recently, a blog post on the Raptive blog: Our take on the future of search: SEO, GEO, and beyond. I don’t want to enter the polemic of SEO vs. GEO vs. LLMEO, because—as we were saying before starting the recording—I think it’s becoming boring and futile.
Instead, you were saying that we should really push readers to understand that there is not just the classic search. People are not searching only on Google’s main search surface anymore. Even inside the Google ecosystem, it could be YouTube, it could be News, it could be Discover with predictive search. And beyond that, people—especially the new generations—are searching on Instagram, TikTok, social media in general, and now ChatGPT, etc., etc.
So, to expand the concept of search, because search behavior is changing. You were talking a lot about news, I believe, in the context of Raptive. Why did you put such a strong emphasis on news? And then I’ll ask you something more.
Tom Critchlow: Yes. Well, the Raptor network, for those who don’t know, is a network of about 6,000 sites we work with: a mix of independent content creators and bloggers, but also independent news and media sites. We definitely skew to the content end of the spectrum.
We don’t really work with e-commerce sites. We don’t really work with government sites or anything like that. So our view of the web is a certain kind of slice.
And I stand by what I said: for most creators and news, and media sites, there’s nothing you should be doing differently today for AI search than for regular search—assuming you’re doing modern SEO best practices. Nothing has really changed. I haven’t seen real evidence—case studies, tactics, strategies—that you should do differently than before.
That doesn’t mean nothing is changing. Everything is changing. AI Overviews, AI Mode, the things people are searching for, how they search, the platforms they’re using, where everything is showing—everything is changing. But in terms of what you do about it, I don’t think much has really changed yet.
Again, we have a fairly opinionated view of the web—heavily skewed to content creators. I think the whole SEO vs. GEO debate is a little reductive. Obviously, it’s a dumb argument. Mostly because we’re thinking of SEO as one thing and GEO as one thing. That hasn’t been true for a long time. SEO has always been different things for different types of sites.
If you’re a platform like Airbnb, what SEO means to you is radically different than if you’re The New York Times. If you’re an e-commerce site, what you think of as SEO is radically different than an independent blogger. It’s a different focus on technical vs. content vs. editorial. Different kinds of search results, different kinds of search features, different kinds of schema—everything’s different under the hood.
And the GEO debate is the same: GEO is not one thing. If you’re a content creator—news, media, blogger—I think GEO is mostly the same as SEO today. Most of the things you do are the same.
Tom Critchlow: If you’re a brand—e-commerce or otherwise—GEO is a little different. And my opinionated take is that increasingly we’re going to see brand reputation—not just classic “brand authority” in a page rank or link-equity sense—is increasingly going to drive how you show up in AI search surfaces.
If people say bad things about your brand—if they reference your brand with “don’t use them for X”—that’s going to have a knock-on effect on how AI search engines feature and talk about your brand.
Again, back to the conversation about Web Guide: the future of SEO—whatever surface you’re looking at (ChatGPT, Google)—is that these platforms need a reason to link to you. It’s not enough anymore to just show up number 1 in a search result . They need a reason. They need to write that little bit of text that says, “You should click on this link because…” And if there isn’t a good because—if you’re just showing up number 1—you’re not going to make it in the future.
That’s much more of a brand reputation, a UGC, consumer sentiment type play. For those kinds of brands, that’s going to become increasingly important. But for news and media sites, that’s much less the deal. It’s much more about the content itself rather than the brand per se.
So yes—everything is changing. But what’s lost in the SEO vs. GEO debate is the nuance that SEO is not one thing, and GEO is not one thing. Depending on your business, your brand, and your industry will dictate how you should adapt to the future.
Reputation & Sentiment: ORM in the LLM era
Gianluca Fiorelli: Yes, and I think that online reputation management—which was another area of SEO—is going to resurface. That’s why I usually recommend that whenever we’re doing—let’s call it keyword research (even if I don’t love the term)—and then clustering, we cluster all the keywords Google suggests and also run sentiment analysis to understand the implied sentiment.
Especially for branded keywords, to understand the implied sentiment—is it positive or negative? Because depending on the cluster, we might see: we are beloved for one type of product, but another product line is considered awful, and people are saying the best about it.
So we can understand—and also inform the client—and say: “You need to do something to win back your customers regarding the quality of this product.”
This isn’t classic SEO as it was—but it’s still SEO to me, because acting on that insight can have a better impact on brand positioning in LLMs, AI search, etc.
Tom Critchlow: But there’s a distinction here, right, Gianluca? This goes back to the conversation we had earlier. As an SEO professional, you can highlight a gap in brand perception—for a certain product, service line, or location. But changing that brand perception? That’s not the SEO team’s job, right?
So this also goes back to the point that SEOs talk a good game about having all the answers, but the people who drive the real outcomes—the real SEO outcomes—are often not the people who work in SEO.
Gianluca Fiorelli: Yes. For instance, earlier I was talking about online reputation management. In the classic ORM—just because it was Google—we were playing only with Google’s algorithm: push the bad results to page two and beyond.
But with LLMs, you cannot hide anything—it’s going to be picked up anyway. You can’t just build an enormous amount of pages saying you’re the best brand in the world. One popular Reddit thread can say the opposite—and that thread will surface, maybe even as the first thing the LLM says about your brand.
Tom Critchlow: Yes, and I think this is, you know, an opinion about where things are going. Just to get technical for a second: we live in a world right now where I perform a prompt or a search phrase, like “SEO agency in New York.”
As I enter that query, the way ChatGPT and Google work today is they do what’s called query fan-out. Under the hood, they execute a bunch of related searches in real time: “best SEO agency in New York,” “e-commerce SEO agency in New York,” “enterprise SEO agency in New York,” “best SEO agency,” and so on.
They assemble those search results, and for each of those search results, they extract the text from the page, and then summarize all of that to generate a response back to you. That’s how things work today.
And we actually know from some of the leaked antitrust trials recently—it was in one of the documents that was unsealed. They have this thing called Fast Search, right? Under the hood, when Google does those query fan-outs, they’re not using the same fully featured search we use as consumers. They’re using Fast Search, which—as the name implies—is faster.
So, they emphasize speed. They need to run those fan-out queries in, you know, <50ms, get results quickly, fetch them, and quickly summarize them, to have any hope of returning an answer in time for AI Overviews or AI Mode. That’s how things work today.
I strongly believe that very soon—maybe Google is already doing this today—when Google crawls your site, they’ll run an evaluation of your site at crawl. So when the query fan-out happens in real time for, “SEO agency NYC,” and they get information about your site or brand, that info will be enriched with a precomputed LLM response and evaluation of whether the business should be trusted, what its authority is, whether there’s bad press, and why it might be relevant for specific kinds of searches.
Google will precompute that using LLMs so that their Fast Search response can grab not just the URL and the text from the page, but also this metadata around how much it’s trusted—like what its kind of PageRank is in the classical world—but also a bunch of evaluations, qualitative evaluations that’s then going to get pumped back into AI Overviews and AI Mode, right?
And I don’t think that’s happening today—we haven’t seen any evidence that it is—but it’s clearly where things are going in the future.
So, back to the point about reputation management and brand perception: these things are going to get precomputed at some point. We don’t know how often; it’s going to be a black box. Google will do this under the hood, and we’ll have no idea how they do it, how often they refresh it, or what sources they use to compute it. But that information is going to radically change the quality and the relevancy of AI search results.
Gianluca Fiorelli: But I can foresee it somehow—it’s speculation, but I can see it as an expansion of the Knowledge Graph concept. I mean, the Knowledge Graph is not just about a brand that has this CEO, producing this product, sold under this brand name, etc.
But adding and connecting things coming from other websites, Reddit, and people’s opinions on this brand (good, bad, medium). I can foresee it going in that direction, because that would be creating a Knowledge Graph based on a knowledge base. And that can be updated, as it was the model. Not as frequently as the classic index, but let’s say, every six months or whenever Google really needs to refresh the data.
Creator Economy in the AI Era
Gianluca Fiorelli: And I want to talk again about what you said—that, at Raptive, most of the people you support are creators and news websites: independent news sites, or blogs, etc. We know these websites can really struggle with AI because, with AI Overviews, we’ve seen it’s not really bringing traffic.
Usually, we see these creators screaming at Google because of the traffic they’re losing. I’m curious—inside Raptive, if you can share—how are you supporting your clients so they’re not so dependent on classic organic traffic?
Tom Critchlow: Yes, there are two things.
First, we advise all our clients to try to diversify away from Google: build your email list, build your social following. These have been best practices for years—and they’re obviously important. However, it’s also true that Google is the number one traffic source by a long way, and there isn’t really a good number two. So, yes—diversify (you should absolutely do that), but it won’t necessarily tell you the whole story.
Second, we also advise our clients—and everyone we work with at Raptive—to block AI training bots. We don’t enforce it, but we recommend it. We believe that over the next X number of months—however it shapes out and when it shapes out—the balance of power between the AI models, Google included, and content creators is going to rebalance a little bit.
You remember when LLMs first came out? It was like: the foundation model can just answer completely. But we realized they had massively high hallucination rates. How do you fix that? You ground it in search—the query fan-out, and so on, that we were just talking about.
So, you need content. You need fresh, authoritative, real-time content across the web to make a good response. If those responses don’t link out—if they don’t send traffic out—that content ecosystem dies. And Google knows this, and ChatGPT knows this. They just haven’t done anything about it yet. But we fundamentally believe that the balance of power has to shift back somehow.
And we believe that if you block the training bots, you gain negotiation power—future bargaining power—when we finally get a seat at the table to talk about whether and how we get compensated.
There’s a ton of work happening in this space. There are startups, and Microsoft just announced an initiative last week. Lots are happening in real time. I don’t think anyone has really figured out—or even knows—where that goes or how that problem gets solved. But the good news is that we know it’s a problem.
And all the AI platforms—Google included—know they need this content ecosystem to exist. If they don’t send traffic out, the content ecosystem dies. They know that, and they’re heavily incentivized to solve it.
Now, whether we end up with as much of the long tail on the web as we used to—I don’t know. Whether it’s still ad-supported—probably, but we don’t know. Lots of open questions remain about how this future state will end up.
But fundamentally, yes, you can try to diversify your traffic away from Google, but for most businesses, that’s a very difficult, multi-year effort. Now, it’s something you should do—I recommend it—but it’s not an easy recommendation. And, again, Google is still the behemoth in the room.
Gianluca Fiorelli: Yes, and related to this, in a short video you did for Raptive, you emphasized the importance of creating value, and this is where the creators are really strong.
I was thinking about a conversation I had with Joanna Lord about the creator economy. Until now, brands and creators usually run on parallel paths: the brand is the cannon, and the creators are the fandom of the brand. They rarely cross paths.
We were foreseeing—at least Joanna was—the need to start making creators and brands work on the same path, not parallel ones. Not just simple co-marketing or influencer marketing events, but really co-creating things together—so brands can absorb part of the fandom, and the fandom stays aligned with the brand’s values, vision, and ownership.
Gianluca Fiorelli: Do you think, considering the kind of creators you work with, this could be part of their future?
Tom Critchlow: I mean, I think this also goes back to something I talked about earlier—SEOs often think they have the ability to influence this stuff. But if you’re an e-commerce brand, you likely have a team managing influencer relationships. You have vendors, analytics, and brand teams doing brand deals.
This is a whole ecosystem and a whole industry around influencer marketing. It’s not new—it’s been around forever, all the way up to celebrity partnerships on TV. If you watch any TV ads these days—and I try to do as little of that as possible—almost every brand has a spokesperson, a face of the brand, whether that’s a mascot, like Geico, or whether it’s like a mezcal tequila with Dwayne Johnson, “The Rock”. Every brand has a face, and I think what Joanna’s sketching out a little bit is that we’re still relatively early in having all the infrastructure to support brand partnerships and influencer work at scale. But that space is more mature than a lot of SEOs give it credit for.
A lot of real brands are spending a ton of money there, and a lot of creators in the Raptive network are getting paid a ton of money there. I think that space will continue to evolve.
And I also feel like the insight Google had a couple of years ago—that people want firsthand perspective and real users’ point of view (one reason Reddit exploded in the SERP)—is the same insight that gives me hope that, as LLMs get better and faster at processing and surfacing information, we’re going to see more of that firsthand perspective.
More creator-led content, more individual voices, more blogs, more editorial content—coming back into these responses. In many ways, an LLM response today is kind of like Google was five or ten years ago: unopinionated, templated, just a wall of text. But we know people want rich results—linking out to a variety of sources, visual, with a firsthand perspective, engaging, and unique. All of that is going to come back—we’re going to return to all those ideas we knew from before.
So, anyway, long-winded way to answer your question: yes, influencer marketing is a massive thing today, and it’s only going to get bigger.
Gianluca Fiorelli: Yes, I was also thinking beyond the classic influencer market—what we could call the small influencer: the real creators who maybe run a YouTube channel with a small but very influential, highly engaged audience.
These are the kinds of influencers—the real creators, in my opinion. An actor isn’t a creator; a celebrity isn’t really a creator. These are the creators I think brands should start building bonds with.
And about what you said in the beginning: we’re still in the middle of the Red Sea. We’ve started the journey; we can’t see the starting coast anymore, but we still don’t see the other coastline yet.
Another piece is advertising. Neither ChatGPT nor the others have really found a way to advertise inside AI search. They’ll have to re-teach users to click a link. Do you have any hypotheses for how that could work?
Tom Critchlow: Well, I mean, we just saw last week OpenAI announce ChatGPT Pulse, which is their equivalent of a Discover feed. I think it’s incredibly obvious how ads could show up in that surface. This is the LLM engine coming to you with recommendations of what to consume—perfect surface area for ads.
So I think both in the intent-driven flow—I put in a query or prompt and get ads— and in a recommended content ad feed, we’ll see ads.
Gianluca Fiorelli: Like advertising inside Discover?
Tom Critchlow: Correct—but more relevant. It won’t just be a link; it’ll be “Here’s a link and here’s why you should click it.” Everything will come with that recommendation layer. And that, I think, is going to be a massive boom—the amount of revenue flowing through advertising against LLM responses is going to be huge. And that’s why it’s Google’s game to lose. Google is still the number one player, well-positioned to win that battle.
Quick Personal Note: “Not American—New Yorker”
Gianluca Fiorelli: Yes. Well, 52 minutes—I know you have to go, but before you do, a quick question about you. I have a curiosity: you’re British, but you’ve been living…how many years in the US?
Tom Critchlow: Fourteen years now.
Gianluca Fiorelli: So, my question is the same one people often ask me: “How much Italian do you still feel, and how much Spanish do you feel now?” For you—how much British do you still feel, and how much American—US American—do you feel after 14 years?
Tom Critchlow: Good question. The quick, simple answer is that I don’t feel very American, but I do feel like a New Yorker.
Gianluca Fiorelli: Okay—that’s a great answer.
Tom Critchlow: Yes.
Gianluca Fiorelli: Okay. Thank you, Tom. It was a real pleasure to have this conversation with you. I’m sure it was for our viewers and listeners on YouTube, Spotify, and Apple Podcasts as well. A great hour of conversation.
Tom Critchlow: Gianluca, thank you so much for having me on. I really appreciate it. It’s always nice to chat and catch up—thank you for having me.
Gianluca Fiorelli: And thank you all for watching and listening. Remember to ring the bell and subscribe so you’ll be notified of new episodes—like this one with Tom Critchlow. Thank you. Bye-bye.
Podcast Host
Gianluca Fiorelli
With almost 20 years of experience in web marketing, Gianluca Fiorelli is a Strategic and International SEO Consultant who helps businesses improve their visibility and performance on organic search. Gianluca collaborated with clients from various industries and regions, such as Glassdoor, Idealista, Rastreator.com, Outsystems, Chess.com, SIXT Ride, Vegetables by Bayer, Visit California, Gamepix, James Edition and many others.
A very active member of the SEO community, Gianluca daily shares his insights and best practices on SEO, content, Search marketing strategy and the evolution of Search on social media channels such as X, Bluesky and LinkedIn and through the blog on his website: IloveSEO.net.
stay in the loop