Welcome to The Search Session. I’m your host, Gianluca Fiorelli. Today’s guest is Grant Simmons, an experienced SEO consultant based in Virginia Beach, USA.
We get into why so many agencies fail to market themselves with the same rigor they'd demand from a client and dig deep into the technical side of AI search: crawlability, site speed, internal linking, and building knowledge graphs that actually hold together.
Grant also walks through the guardrails that keep programmatic SEO from spiraling into thin, penalized content, and explains query drift, the subtle way pages and entire sites drift away from their real topical expertise.
What you’ll learn in this session:
How agencies can adapt to a changing search market: apply the same strategic thinking they offer clients to their own businesses by demonstrating ROI, building a differentiated brand and a clear niche, and promoting their expertise rather than relying on generic pitches.
Why technical SEO remains fundamental in AI search: technical debt and poor implementation can prevent search engines and LLMs from accessing content, making crawlability, rendering, site speed, navigation, and platform choice essential to visibility.
How site architecture and internal linking strengthen expertise and authority: strong internal linking and logical hub-and-spoke organization help search engines understand a site’s structure and important pages, while content gaps and information gain help define how its content should be organized.
How databases and knowledge graphs can become content assets: connecting entities, attributes, relationships, and available data helps create differentiated, comprehensive content that meets users’ information needs and encourages engagement.
How to scale programmatic SEO without sacrificing value: define the constraints of value based on user needs and real engagement, rather than the sheer number of possible page combinations, to determine which programmatic pages are worth building.
What makes complementary taxonomies valuable at the architectural level: relevant facets can create information gain and differentiated discovery paths, but they should be guided by user needs, tested for engagement, and connected through a clear primary crawl path.
How to balance marketing language with the clarity LLMs need: pages can retain creative brand messaging, but they must clearly communicate who the company is, what it offers, which problems it solves, and the purpose of each page.
What query drift is and how to correct it: query drift occurs when a page or site becomes associated with queries, entities, or topics outside its intended focus and can be corrected by clarifying, separating, improving, or removing outlying content to reinforce topical focus for search engines and LLMs.
What is commonly misunderstood about GEO: expensive prompt tracking does not necessarily provide measurable ROI, while content chunking means creating self-contained, unambiguous passages that serve user intent, not arbitrary blocks designed for retrieval.
Join us for a practical conversation shaped by Grant’s decades of experience across agencies, brands, and search technologies.
Topics covered: agency positioning · technical SEO · site architecture · internal linking · complementary taxonomies · knowledge graphs · data-driven content · programmatic SEO · information gain · content clarity · query drift · content chunking · AI search
About the Guest
Grant Simmons
Director of SEO, Content, and GEO Strategist at Fiat Growth
Grant Simmons is an SEO and GEO strategist with more than 35 years of marketing experience and a career in search dating back to 1996. He specializes in technical SEO, content strategy, site architecture, entity optimization, and connecting visibility across traditional and AI search to measurable ROI.
He founded Simmonet Marketing in 2001, providing SEO and GEO strategies that support organic growth. That same year, he began consulting through Brilliant Strategy, which specializes in SEO and content strategy. Since 2024, he has led SEO and GEO at Fiat Growth, a growth consultancy for emerging brands.
Grant regularly speaks at industry conferences and shares practical insights on search visibility, content, and AI through his LinkedIn articles and blog.
Transcript
Full conversation between Gianluca Fiorelli and Grant Simmons.
Gianluca Fiorelli: Hi, I'm Gianluca Fiorelli, and welcome back to The Search Session. It seems like a rhetorical figure that I already used a few times now, but the person that we are going to have as a guest is someone who substantially lived through all the evolution of SEO when SEO was substantially a very tiny thing and made on Yahoo, AltaVista, and so on, and is still very active and very proactive now in the age of AI search. For this person, this statement is true. It is 100, 200, or 1,000 percent true.
This person is, as his LinkedIn bio says, “An experienced search consultant for brands seeking positive ROI on their organic investment.” It's interesting that in his LinkedIn profile, he also says that he's "ROI-minded," so return on investment is his most important focus when he works as a consultant or as a director of SEO, content, and GEO strategies for Fiat Group.
You probably know him apart because, as soon as you see his face, you will recognize him. You may also know him as an advisor and evangelist for Waikay. This person, our wonderful guest today, is Grant Simmons.
Hey, Grant. How are you doing?
Grant Simmons: I'm great. That's got to be one of the best intros I've ever had in the way of saying I'm old.
Gianluca Fiorelli: Well.
Grant Simmons: I've been around since the beginning.
Gianluca Fiorelli: I have to sell my guests.
Grant Simmons: No, thank you. I appreciate that.
Gianluca Fiorelli: So, how are things going? For context, this is a live recording, so no cuts. It's like a live stream, but we are recording this episode. Today is the 10th of August, so what I'm going to say may already be in the past when the episode is streamed live, but I know that you are going to return to speaking soon at BrightonSEO San Diego.
Grant Simmons: That is correct. And I guess “return”—I've been speaking at a few places this year—but yes, I'm going to be back in San Diego at BrightonSEO. I'm really looking forward to that. I'm going to be talking about entities and some other stuff that's kind of cool and timely.
Gianluca Fiorelli: Surely. And talking about entity search and so on, how is SEO treating you lately? Because these 24 months of total frenzy remind me a lot of the very beginning of SEO.
Grant Simmons: Yes, "frenzy" is a really good word. I think confusion, giving yourselves 10 lashes. Fundamentally, there are a lot of things that people aren't thinking of that go back to the beginning of SEO. Just keeping things clear and clean. Clarity is really cool.
But yes, it's treating me very well overall. I must admit to being one of the saner heads, I hope, just pitching to both clients and on stage that the fundamentals are still important. And, to clients, the fact that if we don't have those fundamentals in place, then building for SEO, GEO, or any kind of surface is going to be challenging.
So, it's been really good. I'm having the best year I've ever had, which is great from the standpoint of keeping busy, keeping people happy, and actually enjoying life. That's all we can do these days.
Gianluca Fiorelli: Yes, and I'm very glad to hear it because it has also been a really good 24 months for me. This interruption of AI in search is substantially also the reason why The Search Session exists. Maybe it wouldn't have existed before, but it gave me the reason to create this video podcast.
Why Agencies Struggle to Adapt and Pitch Value
Gianluca Fiorelli: On the other side, in private email groups, private Facebook groups, and private WhatsApp groups, it is not uncommon to see, especially on the agency side, how many agencies are really struggling to market themselves in this change of era. Maybe it is because many companies of any size, but especially medium-sized to enterprise-sized companies, are moving a lot of things in-house because AI is considered a good substitute for many things that were previously delegated to agencies, for instance.
Also, considering your past experience as an agency owner, what is your opinion about this pivotal moment we are seeing for agencies? From your personal history and perspective, and from your current success, what kind of recommendation would you give to agencies or even consultants who are really struggling to maintain their projections to clients?
Grant Simmons: I think there are a couple of things to slice and dice there. One thing is it's really ironic that agencies that sell marketing services to clients are not able to put the same lens on themselves and look at the same things we would do, which is, "How are you differentiated? Have you built a brand? Are you clearly communicating the products and services you offer?”
And I'm really glad that you read my LinkedIn bio, but the idea of ROI or return on effort is really key. More often than not, agencies pitch themselves in more of a generic box: “We improve traffic,” or, “We deliver 10 pieces of content a month,” without tying that back to real ROI. So I think the challenge is kind of self-perpetuating if agencies don't change their mindset to put the same lens on their own business as they do with clients.
Number two is not jumping on the bandwagon and saying, “Now we're a GEO agency,” when SEO is still a fundamental discipline that should be available to clients. That should help. Content strategy should be part of this.
I think what has happened in the past with SEO is that it became more commoditized. Agencies became more commoditized in both the way they pitched and the way they delivered. If we don't take a strategy-first approach, then everything else we do has no real measure and no real understanding of where the return on effort and return on investment come from.
To me, agencies should take a look at themselves with the same lens that they use to look at clients. Try not to commoditize what you offer, whether that's SEO, GEO services, content, landing pages, or whatever it might be. Really try to create a differentiated message, a differentiated brand, and some kind of niche that you are absolutely expert in.
That's the kind of advice I would give, and I only give it because I've followed it myself. When I had my agency, which is now 25 years ago—is that right, my math? Yes, 25 years ago—I started it by focusing on a few main areas and then grew the agency from there. It doesn't mean that you have to say no to new business. It's just about being selective about the business you take and selective about how you present the businesses, clients, or work you do.
So I think that's where I come from. Hopefully, that made sense and connected with a few people out there.
Gianluca Fiorelli: Yes, indeed. I think you were really correct in saying that sometimes the problem—and I see it with myself too, for instance, with my very horrible personal website—is the classic thing where I say, “Well, I'm a consultant. My website is the last thing people are going to see because they are going to see me in other places and so on.”
So I always postpone it. I'm also guilty of procrastination syndrome, so I procrastinate when it comes to correcting and updating my own website. But this is even worse if it happens to an agency, for instance.
What you said matters to me: you’re not doing for yourself what you appreciate in your clients. You're not branding yourself as an agency. You're not branding your people as brand ambassadors for your agency.
This is something that I also notice. Usually, the people saying that they are struggling are people I may know, but I don't see them sharing stuff. I know there is this sort of dichotomy and stereotype that says the people spending a lot of time on social media promoting things or speaking at conferences are often not the ones who are really doing the work, which is not true.
First, there are exceptions, obviously. But second, if you are not showing off your expertise, you're just depending on clients’ word of mouth, ads, or cold pitches to new clients.
Grant Simmons: A lot of things I hear are around, “Well, I don't have time to do that,” or, “That costs money.” Every time you approach a client, you are asking for some kind of value exchange: I will do this work, I will get these results, and you will pay me this. All right?
So there’s a value that we recognize as agencies or consultants. There’s a value because we want to charge for our time. We don’t look at it the same way for ourselves. We expect to get results without necessarily investing time, energy, resources, or money into actually promoting ourselves.
And look, there are some people who are well known enough in the industry that people knock on their doors, and fortunately, some of the time, that’s me. But that is purely from—they say an overnight success took 30 years to get there. I first worked in marketing in 1980-something. And I’ve been building a résumé, experience, and everything else since then. The first time I worked at an agency in Los Angeles was in 1994. That was not my first marketing job.
I understand there’s always a value exchange that’s really key. There’s a differentiation that the client or prospect has to see and hear, and the only way to do that is to spend some of your own precious time getting yourself out there and sometimes spending a couple of bucks to promote something.
I’ve done both over the last few years. I went out by myself again in 2022 and focused on promoting but not pitching. To your point, and I love what you’re doing, Gianluca; essentially, you are sharing really valuable content that doesn’t necessarily say, “You have to hire me.” What it says is, “I know my stuff, and if it connects with your need, then maybe you should hire me.”
I think too many people pitch, and not enough people promote their expertise. Those are two very separate, different things. Hopefully, as an industry, we can come back and realize that.
Gianluca Fiorelli: Hopefully, yes. Thank you anyway for the kind words.
Now let’s move on from the agency to the actual work. You were saying, okay, you promote your GEO and your AEO, but SEO is still fundamental to everything we are doing now, including AI search.
JavaScript, Site Speed, and the Technical Debt Problem
Gianluca Fiorelli: I wanted to talk about technical debt. The technical side of SEO was very important once and is still very important for classic SEO. But sometimes it seems to me that, for technical SEO for AI search, it’s somehow dumbed down. Yes, it’s mentioned, but usually for just one thing, which is JavaScript, and not for many other things that really deserve attention.
My question is, how much is technical debt—which is very common, especially in enterprise—really a problem that we, as consultants or agencies, must put very clearly in front of the client? Or, if we are in-house, in front of our stakeholders inside the company?
Grant Simmons: There’s something I’ve been saying for a long time, which is, if you can’t be indexed, you can’t be cited or show up in results. Above that, if you can’t be crawled, you can’t be indexed, cited, or show up in results. Now, that’s simplifying everything.
You mentioned JavaScript: it’s still a massive challenge because, a lot of the time, development drives decisions around serving content or serving sites.
For example, some of my current clients have problems with JavaScript crawling. They have problems with rendering. They have problems even showing up consistently across pages within their sites. These are fundamental technology challenges that go back to the days of technical SEO. It’s still a problem.
Where it gets, I think, not so much complicated but more important, is that site speed is still a major problem. Being able to crawl. Google has given a lot of leeway to sites and site speed, and they’ve been pitching it for a long time. I’d say that well-promoted or branded sites can get away with a lot.
LLMs don’t have the same leeway around things like your page being slow or not loading components on a page in a way that they can read. They’re just going to not cite you or give up. They don’t have the same resources as Google or even Bing.
So I think the key thing there is that technical SEO is really, really, really important. By technical SEO, we mean how you serve and provide content. No matter what surface it’s going to end up on, it has to be legible, formatted, and readable. All those things rely on that technical infrastructure to make sure that those fundamentals of “Is it clear? Is it consistent? Is it cool?” show up.
Even today, it is so easy to publish static HTML pages. I don’t know why more people don’t do that for key content. Maybe that’s an obnoxious way of looking at it.
Gianluca Fiorelli: Because it’s not fancy. We would need Jono Alderson in this conversation to demonstrate to developers that you can do hundreds of things with just one line of CSS instead of 200 lines of JavaScript and have fancy things that are usually delegated to JavaScript.
Grant Simmons: Yes. A lot of the time, what I hear is that it’s all about the dynamic injection of content, or there’s some kind of dynamic component, so they’re using JavaScript to inject it.
But you’re 100% right. You can use modern CSS to do almost anything that JavaScript can do. Being able to ensure that content is readable later on a page, on page load—you and I remember that as something we were talking about maybe 25 years ago.
Martin Splitt was talking about it when JavaScript frameworks were coming out. That was in 2016, I think, when he was at BrightonSEO talking.
Gianluca Fiorelli: And before, it was the Flash website era.
Grant Simmons: Flash websites. We're from the era of Flash. That really makes me feel old. But that's what I'm saying. To put it succinctly, the technical SEO choice of platforms is really important. I'm not going to attack any platform in particular, but Framer, all right?
I have a client with a Framer website that has dropdown navigation. The dropdown navigation is loaded on rollover or on click. Therefore, it's not visible on the page when it loads. Everything that relies on the top navigation is not found, so you have to use workarounds.
The workarounds that Framer suggests are creating an HTML sitemap, putting the links in the footer, or putting the links in the content. To me, that's a platform working against itself. And I'm not calling them out. There are lots of platforms that have these inherent workarounds that we shouldn't have in 2026.
Technical SEO is still important. Understanding the choice of platforms is important. Don't leave it to marketers to make everything headless because it's kind of cool. That doesn't mean it's not good, it isn't an idea, or it can't work, but we have to make sure that the content can be crawled, indexed, cataloged, understood, and presented well. So that's fundamental.
What does technical SEO look like in the age of AI?
Listen to more expert takes from the Search Session:
Will Kennard: technical SEO as creative strategy, where experimentation and problem-solving drive innovation.
Jamie Indigo: technical SEO for AI search, from machine-readable signals to optimization in LLM-driven discovery.
Jono Alderson: technical SEO from first principles, questioning how much of our day-to-day work is actually moving the needle.
Alfonso Moure: technical SEO for the agentic web, from AI crawlers and log-file analysis to machine-readable signals and faster server responses.
Site Architecture and Strategic Internal Linking
Gianluca Fiorelli: And there is an aspect that can be considered both technical and content, and it's architecture. Architecture, technically, is substantially how you plan internal linking in a consistent and logical way, based on a content strategy you have in mind too.
Then it's content because it's the skeleton around which you establish your areas of expertise with clusters, specific landing pages, and so on. Architecture was obviously important even before, but how much more important has it become now for you?
Grant Simmons: Just for transparency, I work as an ambassador for Waikay and InLinks. InLinks does internal linking, and I joined that company because internal linking was something I was pitching a long, long time ago.
Internal linking is one of the few things we have complete control over to reinforce expertise and authority and to reinforce important pages on our site. So internal linking is unbelievably important.
There are different types of internal linking, obviously: navigational links versus links in the content and anything else like that. There are differences based on the Reasonable Surfer model, where links that are more likely to be clicked are generally given a higher weight or value.
Now, that's just a patent, but it makes sense. It makes logical sense that if a link is hidden, it probably isn't going to have as much value to a user as one that's within content, that is topically relevant, contextual, and leads to a logical page that expands the conversation or topic, or is a logical connection to that main page and topic. So, internal linking, 100%.
You also talked about architecture. You talk about the hub and spoke. You can talk about silos. You can talk about all this stuff, but logical organization is important.
I always come back to this metaphor: Google wants to be the world's biggest library, and you don't walk into a library and see a pile of books on the floor. You see organized books along shelves in alphabetical order, using the Dewey Decimal System in the old days. Every shelf shows exactly what's on that shelf, and you know how to find stuff easily.
For me, Google has the same kind of organizational need. They want to understand how a site is structured. They want to understand the important parts of a site, generally through internal linking, contextual anchor text, and things like that. Architecture is really key.
You also touched upon one other thing. You talked about content strategy. You're right: a content strategy can both hang on the bones of an architecture and define the architecture.
There are a few ways I look at content strategies around gaps. What exists on competitor sites that we don't have on our site or our clients' sites? The second thing is information gain. I know that's used in many different ways, but information gain, to me, is what doesn't exist out there on the web or is only loosely covered. It may not be covered by competitors, but it is something that adds value, differentiation, and normally some kind of stickiness or appreciation of your expertise.
That's normally your own data visualizations or unique content based on your own views, viewpoint, and experience—that whole E-E-A-T type of stuff. Look at existing gaps and fill those, while also trying to find information gain: information that either doesn't exist or exists only in a very thin way across the web.
To me, that helps define and redefine an architecture, redefine how you organize stuff, and fill in the gaps of what might be missing.
Knowledge Graphs vs. Databases for Search & LLMs
Gianluca Fiorelli: Totally true. And talking about content as data, I have a question, and I want to pick your brain. It's an open question.
Grant Simmons: You don't want to be in my mind. You don’t want to be there.
Gianluca Fiorelli: Well, it's truly a curiosity I have because it's something that I also want to think more about. We always talk about optimizing content, optimizing the technical side of our website, and optimizing our visibility on third-party websites, etc. There are a few people—and I think the more semantic, knowledge graph-aware people—who also talk about databases.
The database can be the physical database from which our website pulls all the data that populates the front end. But it can also be considered our knowledge base, which we must create and maintain to eventually make our content more meaningful, consistent, and connected.
Not many people are talking about how to optimize a database, which also has strong implications for page experience, velocity, page speed, and so on. What about creating the database as a content asset for your website and your presence online? What are your thoughts about these topics?
Grant Simmons: First off, I'm a massive knowledge graph proponent. When I'm building out a site architecture, I'm looking at entity representation within pages. I'm looking at disambiguating through entities within the graph.
We can get really deep and geeky here, but essentially, the knowledge graph that you build should completely cover the topics you are an expert in. If you have connections between entities and entity attributes, you can tie that into content in the same way.
A few years ago, I used the hierarchy of Star Trek because that connects with geeks quite well. If you have the Enterprise, you have all the crew members, the attributes of the crew members, the relationships between the crew members, and the episodes that each crew member is in.
You create this amazingly rich knowledge graph that is very similar to how Wikidata or Wikipedia works, where there are connections between your entities, with the relationships and attributes stated. You end up being able to create pages that actually have value because they're differentiated and connected in the right way.
So, 100%, build a knowledge graph. I probably don't equate that to a database in the same way that you said, but that's not to say it can't be. I look at it as a key component of building out a complete site graph and, if I'm planning from scratch, making sure I'm covering all topics.
As far as the database, data is amazingly available right now. Whether it's your own data, government-based data, data from studies, or anything else, the difference in using a database to create content is whether it connects with the user's need for information and the search engine's or LLM's need for information.
Generally, they don't want to put a graph chart in LLM results right now, something dynamic, or anything else. But we have the capability, as SEOs, content strategists, or whatever else, to create data-driven content. We have the responsibility to create something that people are going to use and engage with.
So, to the two points there: data-driven websites are really key, and knowledge graph-driven websites are really key. Connecting those can create additional content strategy enhancements, if I want to use a better phrase, but it has to be something valuable to the end user.
At the end of the day, we know there is an engagement metric that Google uses to see whether there is satisfaction based on the query and everything else. We want to make sure we're sending the right satisfaction signals, whether that's dwell time, query modification, or anything else.
We want to make sure we're sending the right signals to Google that the page the user lands on—and, down the road, this will be the same for LLMs or AI Overviews—is actually providing a satisfactory answer that the user will engage with.
The user will build trust in both the engine or surface that sent them there and the resource itself. So that's a long way around saying, "Yes, I agree.”
Programmatic SEO: Scale Without Losing Value
Gianluca Fiorelli: Okay. Can the answer you just gave also provide useful guidelines for people who have websites with strong databases and tend to create overly scaled programmatic SEO on top of them?
We have seen many cases of websites being penalized because of scaled content resulting from programmatic SEO pushed to the extreme. Not just because you can connect all that data in thousands of different ways, but you are obliged to or should create all those connections live on the front end.
So, is the usefulness of a connection or a programmatic evolution that you may want to pursue the first and most important constraint when deciding whether to do something or to push programmatic SEO too far? Am I right?
Grant Simmons: I think programmatic SEO can work, and it has been proven to work, but it has to have guardrails. I was the VP of Search at Homes.com. You can imagine that we had 100 million US properties listed on Homes.com and around three million location pages.
What you've got to decide at the outset is what is a constraint of value? Where is the point of value to the user that's going to constrain how much we scale? Is it a neighborhood, ZIP code, point of interest, or city? How often do those actual results pages have results?
If you have pages that only ever have one result, why would you build a programmatic page for that? You’d probably build out a page because it has no value to users, unless it is really distinct.
I'll give you a silly example. I built a website called Frypedia that's essentially about French fries in the US and whether they are vegan, vegetarian, gluten-free, and things like that. You can slice a fry so many times, but ultimately it just ends up as a slither.
You've got to work out what people engage with. After launching, I found that people wanted to know whether Burger King French fries were vegetarian. That's a big thing. So then I look at that and say, “How do I take Burger King and fries and build out more content around that?”
That's obviously interesting to people, but I'm letting people lead the programmatic conversation. What should I build out based on engagement, rather than what should I build to have the biggest site in the world?
I think that's the constraint people don't consider. When they look at programmatic SEO, they work out that X × Y × Z equals 10 million pages. What they don't say is, “If I do X × Y and people are actually engaging with any of those Y pages, then build out Z.”
That can be something that's actually proven because users are looking at those things to solve their problems. So, programmatic SEO is good. Overscaling is bad. Look at how you avoid slicing the grape so much that there's no grape left for people to taste. I'm mixing metaphors all over the place, but hopefully that makes sense.
Gianluca Fiorelli: I like metaphors, and I think that metaphors, if used well, are an easy way to make people understand things. Returning to architecture, programmatic SEO, and information gain.
Grant Simmons: Yes.
Creating Architectural Taxonomies for User Experience
Gianluca Fiorelli: Could a correct use of programmatic be a way to create complementary taxonomies that are pure information gain?
I'm thinking, for instance, of something Airbnb did many years ago. It had all the rooms, apartments, and so on, and then created navigation based on characteristics such as apartments, tree houses, igloos, and all these unusual classifications of apartments, houses, villas, and so on. This gave it a unique way to navigate and connect the place where you want to spend your vacation with an experience.
Grant Simmons: Sure.
Gianluca Fiorelli: Could this be a way to introduce information gain at the architectural level? Many times, information gain is considered only in relation to a single piece of content.
Grant Simmons: The answer is yes, but once again, it requires a value-based approach.
Gianluca Fiorelli: Obviously.
Grant Simmons: There are a few different ways of defining value. There is value based on search volume, so traditional keyword search volume, which is owned by a few companies. Google doesn't do a very good job.
It can also be defined by the impressions you see in Google Search Console. These may reveal topics you haven't covered but that Google is mapping to your page, which might give you ideas for expanding into other things.
Then there is promotion discovery, the idea—and you and I talked briefly about critical thinking—of putting myself in the user's shoes and asking, “What would be unique for me?”
An igloo is a classic case. You don't have an igloo in Manhattan, but if you're staying in the Arctic Circle, “igloo” may be a good facet to have. You've got to think of these relevant facets that you can surface from a differentiation standpoint. They may have zero search volume and probably zero search intent. These are the kinds of zero-click keywords that people may eventually search for or that you get mapped to because they are unique. Maybe one in a thousand or one in a million people searches for it, but you are the only result that shows up.
So yes, but you've got to create a page that has value. If you create an igloo page for Airbnb, it might list all the igloo Airbnbs in the Arctic Circle or wherever they may be, but you should also create some value on that page. Talk about why they're warm. Talk about why the Inuit built igloos as opposed to square buildings. Present these things as “Did you know?” facts.
You add information that isn't unique to the internet but is unique in the context of a place where someone may want to stay.
So yes, having multiple ways of reaching the endpoint is key when building out architecture, but having one primary crawl path to that endpoint is also important. It's important from an internal linking and breadcrumb standpoint, and all that stuff.
To answer your question, yes, you can come up with things that no one will ever search for or visit. But you can also think intelligently and critically about things that are within the realm of possibility, build them out, see what Google thinks, and then double down if they work.
We did this with apartments and Airbnb and Homes.com. We tested it, and if it didn't work, we got rid of it because having extra pages isn't valuable. Having the right pages with positive satisfaction signals is really key. You can self-correct instead of expanding infinitely.
Communicating with Clarity: Marketing Copy vs. LLM Understanding
Gianluca Fiorelli: And talking about language, there are obviously LLMs and conversational language, but even before, when AI wasn't really a thing, Google was already being pushed by…
Grant Simmons: Hummingbird and things like that as well.
Gianluca Fiorelli: Yes, Hummingbird and RankBrain. Then came BERT, MUM, and so on—natural language processing algorithms.
It's interesting, the definition that you gave, which is very similar to something I also say: we sometimes attribute too much intelligence to what LLMs are actually doing. That's especially true now because we can see how convincing LLMs are in their prose when writing and answering.
But if you really think about them, they have the knowledge of, as you say, a blind five-year-old child.
Grant Simmons: Not my term.
Gianluca Fiorelli: Maybe now six with every new model. There are still things that all the models, from Gemini to Perplexity, Claude, and also less-known ones like Mistral, are really struggling with.
One is surely rhetoric. They don't understand rhetoric, which is weird because rhetoric is based on rules, and they could very easily be instructed, or we could train LLMs very easily, about rhetorical figures like synecdoche, allegory, or something like this.
Then there is the classic dichotomy that we usually have to discuss with clients. The marketing team of a client or our company uses this very marketing-oriented language, which is based substantially on ambiguities because it uses a lot of allegories, metaphors, and decorative descriptions. That can be problematic for an AI model to understand and disambiguate.
In this sort of arbitration, we have to say, “Okay, you can use this kind of communication, but remember that AI also needs clarity above all.” How do you usually manage this kind of arbitration with your clients?
Grant Simmons: Sure. I can give a really good example. I was doing training for a global, multinational company in Florida. I put a phrase from its website on the screen and asked, “What does this company do?”
It was funny that many of the people in the room, who were looking at their own company's tagline, or not the tagline, but what was on the website, didn't know what the company did. I then showed them that it was actually their own webpage.
I think that's the idea. We're talking about blind five-year-olds, and AJ Kohn, thank you for that. But if we're talking about being unambiguous, a lot of clients think that means they can't use marketing speak at all.
My way of looking at it is that it just shouldn't be the most prominent element on a page, something that is not unambiguous or something that doesn't state clearly who you are and what you do.
With Waikay, we did a study where we looked at what LLMs were actually looking at. Homepages and About pages represented about 7% of what LLMs looked at. If you show a client their About page, it normally has their mission statement or leadership, but it doesn't explicitly say who they are, what they do, or what products they offer. That is fundamentally a challenge for LLMs, which don't understand some of that context.
I think you're right. LLMs are getting much better. I'm often very surprised that when I use some syntax, lexicon, or metaphors, LLMs understand them now. But there are still a lot of apologies when they don't.
It's kind of funny that we treat these LLMs as equal conversational partners, but they're really not. When we're clearer, the answers we get back are clearer. When we're clearer in how we present information, the answers come back clearer as well.
Using Entity Maps to Strip Away Ambiguity
Grant Simmons: I know you have been playing around with Entity Map, entitymap.org but that provides what I call a meaning layer or knowledge layer that takes what a page is about and makes it contextually aware for LLMs. It clarifies what a page is about, as opposed to all the marketing copy.
It really gets down to the root, the meat and potatoes, as I say with my English roots, and links to relevant knowledge sources, such as Wikipedia articles or Wikidata, that clarify the focus of that page.
When you strip away all the marketing speak, are you actually saying who you are and what you do throughout the page? Are you saying what your product does? Are you saying which problems it solves?
LLMs require more clarity and less ambiguity. One way of providing that is by improving on-page content. Another is by creating an entity map so people and search engines can see exactly what a page is about.
There are all these emerging technologies and standards, but ultimately, be clear and concise. Make sure that if you strip away everything, you can still understand who you are and what you do, what your product does, or what a particular page is about.
This goes back to the primary content on a page. It should clearly state and focus on the job to be done that any user who reaches that page wants to do, solve, learn, or understand.
Looking at it through that lens goes back to the Quality Rater Guidelines, where they talk about the primary focus of the page and the primary job to be done.
Gianluca Fiorelli: Yes, its beneficial purpose.
Grant Simmons: That's right. What is the purpose? That's exactly right. I always use metaphors or analogies. If you have a restaurant, the purpose of going to that restaurant is to eat. If it's an Italian restaurant, you want to go there and get Italian food.
Now imagine you go into an Italian restaurant and there's no pizza on the menu. Suddenly, you say, “Hold on. The purpose of this restaurant is to provide Italian food, but it doesn't really provide Italian food as a whole.”
That’s the same way I look at going to a webpage. If you know what you want to do on that webpage, such as engaging, learning, converting, or buying, but you can't easily find that or understand that it's the right place, you're going to send all the wrong signals back to search engines and LLMs.
We want to make sure you can fully satisfy the query, prompt, and intent when you reach a page.
By the way, I'm not known for being unverbose, Gianluca. So, in a nutshell, know what the page is about. It has to have a purpose to exist, and if it doesn't, it shouldn't exist. Done.
Diagnosing and Resolving Query Drift
Gianluca Fiorelli: Indeed. When I was preparing this episode, I found you talking about query drift many times.
Grant Simmons: Yes.
Gianluca Fiorelli: What do you mean by query drift? If it's a problem, how do you diagnose and resolve it?
Grant Simmons: Let's look at query drift in a simple way. If we believe a page is focused on, let's say, pizza, anything on the page that isn't about pizza is a distraction. When Google maps your page to queries, you can see clusters around pizza in Google Search Console, in this particular case, and maybe there are subclusters around cheese, ingredients, or diet types.
But maybe you've talked about pepper, and Google is mapping you to pepper spray because you've talked about spraying pepper on your pizza. That's an example.
Query drift is where we identify the topics, subtopics, entities, and queries at its root that aren’t associated with a page or we don't want to be associated with a page. We can assess them and ask whether another page deserves to exist specifically about that topic or whether it's something we should clarify or disambiguate. Rather than saying “spray pepper on,” we could say “sprinkle pepper on” to avoid ambiguity around pepper spray. That's an example I just pulled out of my head.
Those are the things we have to look at. I developed a tool with my mate Richard Wong, called QueryDrift. It analyzes your site. It's free until you want to get a strategy.
Essentially, we crawl a part of your site and we look at pages around topics and entities. We build a map of the page, and you can see how clusters appear throughout the site and at the page level as well.
You want to find out whether you're mapping to the wrong queries because, if you are a topical expert, you should be known for that topic. Anything that's an outlier should either have its own page or shouldn't exist. Resolving those ambiguities is key to maintaining focus. Query drift is about that, and it's also a technical term.
Gianluca Fiorelli: I think it could also be expanded to the website level. I remember the case of a hotel website. The hotel also had a wonderful spa, but because of how its content strategy had been developed, it started creating a huge quantity of content about spas and wellness.
Google decided that it wasn't a hotel website but a wellness website. It was always appearing for wellness terminology and queries, which we could also see in the metrics in Search Console at the query level.
Could query drift also be expanded to topical drift?
Grant Simmons: Exactly, at the site level as well. We analyze it at the site level first and then at the topic level to understand whether content should be improved, culled, or diversified by breaking it into separate pages or creating a new page. It may also need to go away if it's a complete outlier.
That supports the whole idea of culling or improving content for topical authority.
I can build on the same example. There was a company that provided different generic drugs, and it found that one of its best-performing articles or surveys was about drugs for impotence.
I'm only saying this to make it fun. Essentially, it started creating a lot of content around sex and sexual topics. We ran it through InLinks and found that its strongest topical presence was around sex.
I'm using this as an example, not to be naughty, but to show that it completely changed what the site was mapped to for all the other drugs it wanted to appear for. That kind of drift toward something that isn't relevant to the core purpose of the site can cause problems.
Gianluca Fiorelli: And it may also risk being filtered by a SafeSearch algorithm.
Grant Simmons: That's right. Medicine was still within that Your Money or Your Life area, but you're 100% right. It's the difference between letting marketers decide what might be really good because it performs well but sits outside the site's core concepts and letting SEOs collaborate on a content strategy where we draw boundaries.
If we're going to create something for Apartments.com, then we create something related to apartments, such as pets and apartments, rather than something that's entirely about dogs. If I'm going to create something about dogs, I might cover the sizes of dogs that apartments accept. That ties into the core concept without going off into how many different breeds there are.
We often see marketers look at an area and say, “There's lots of opportunity there. Let's build that out,” without considering how it distracts from the site's core topical authority and topical alignment.
Query drift is about bringing you back to your topical expertise instead of creating content within pages or across the site that confuses search engines. LLMs are similar, but not quite the same. They look more at the page level than the site level. That's why I say keeping pages focused and tight is really important for LLMs.
Debunking AI Search Myths and Pet Peeves
Gianluca Fiorelli: Indeed. One last question about LLMs and AI search. After 24 months or more of AI search being discussed and written about everywhere, what is the most annoying GEO or AI search myth that really makes your eyes roll?
Grant Simmons: Maybe it's not so much a myth, but it is the idea of spend versus value. There's massive investment going on without measuring the actual outcome. That annoys me immensely.
If you can't see, measure, or monitor, or give some kind of ROI on the amount of money people are investing, that's a problem. There are platforms charging 10 or 15 grand a month for what is basically prompt tracking. That annoys me immensely because you don't need to track a million prompts to understand directionally where a site, page, or topic is focused. So that's number one.
As for myths, there are so many. "Chunking" is a word I've used, and it really means making sure you're using triples or something self-contained, so it makes logical sense and is unambiguous. It is not about creating a chunk of 100 words and placing it on a page without any other context. That's not how retrieval works, how content should work, or how users look at content.
Chunking is about making sure it's an unambiguous piece of content that directly targets a particular user intent, rather than looking at it as something that's just going to be pulled out. So the use of the word “chunking” may be the thing that annoys me.
Gianluca Fiorelli: If we wanted to use a nerdier word, it would be making the passage monosemantic.
Grant Simmons: Passage ranking has been a thing for a while. We suddenly change the terminology for things because we think it makes them sound smarter, but it doesn't. Passage ranking has been around for a long time.
Expand your understanding of entities, knowledge graphs, and site structure with these episodes of The Search Session:
Dixon Jones examines how entities, internal links, and topic hubs work together to communicate a website’s structure and areas of authority.
Beatrice Gamba explains how businesses can use knowledge graphs and ontologies to connect fragmented data, define their core entities, and create a shared semantic layer.
Lazarina Stoy connects entity relationships and knowledge graphs with topical coverage, query fan-out, and visibility across traditional and AI search.
Maintaining Passion Across 25+ Years in SEO
Gianluca Fiorelli: To conclude, a question about you. In my introduction, I said that you are one of the few people still working after many years. You've worked in the industry for 25 years, if I'm not wrong.
Grant Simmons: A bit more than that, but let's say I started as a 10-year-old. How's that?
Gianluca Fiorelli: What is your secret to maintaining your enthusiasm for this industry and this kind of work for so many years?
Grant Simmons: I'll give you two answers. First, I'm enthusiastic about everything. I have a zest for life. I'm very lucky. During COVID, when we were asked to write a few articles for the industry, for Search Engine Land, I wrote about the attitude of gratitude. It was about the idea that if you're grateful for what you have, then what you have is enough.
So, number one, I have a naturally positive outlook on life, which helps me. Number two, this industry has been amazingly good to me. It's been good to me in terms of making friends and having some brilliant conversations late at night, generally over too many beers. It's been good to me financially, allowing me to build out a few agencies, help a few agencies, keep clients happy, and generate revenue.
That enthusiasm comes from loving what I do. I love SEO, Gianluca. I love what I do. I love the client interactions, even though not everything is always positive, and I have a very positive outlook on life because I have a wonderful family. I have a fantastic wife and great kids.
I'm fortunate enough, and this is going to blow up your whole thing, to live in the US, which has given me all these opportunities. Make SEO great again. I'm fortunate enough to do that and make a living doing what I love.
Someone said, “If you love what you do, you'll never work a day in your life,” because it's good. I hope that gives you the idea.
For everyone listening, and hopefully there are a few people engaging here, anyone I've interacted with, thank you. Whether I've been speaking from a stage, answering a question afterward, or you've come up to me and shared what you're doing, I appreciate that. I really am grateful.
This industry was originally about sharing information altruistically, and I'm still in that place. You've got to share what you know and what you find without expecting anything in return. This industry has always been magnificent about that. We can omit the snake oil salesmen from that declaration.
I think the core SEOs and GEOs who are trying to do great work every day are doing that great work. I'm really glad and fortunate to be part of this industry.
Gianluca Fiorelli: With this wonderful chant and poem about the search and SEO industry, let's finish here, Grant. One hour is already a long time, even for the most enthusiastic listeners of our video podcast.
Thank you, Grant, for being my guest and a guest of Advanced Web Ranking for The Search Session. I hope to see you soon in real life too.
Grant Simmons: Hopefully, I'll see you at Brighton.
Gianluca Fiorelli: Let's see if I come to the Brighton UK edition. I surely won't be able to go to the San Diego one.
Grant Simmons: My pleasure either way. Thank you again for having me on. Cheers.
Gianluca Fiorelli: And thanks to all of you who had the patience and enthusiasm, like Grant, to stay with us until now. Let me do the YouTuber part. Subscribe to the channel. Give this episode a like. It really deserves it. Ring the bell so you will be notified whenever a new episode is uploaded and released. Thank you, and 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.
