Thanks for joining me on The Search Session. I’m Gianluca Fiorelli, and in this episode, I’m talking with Kevin Indig, a growth advisor with deep experience in SEO. This episode looks at how AI is changing search, SEO, and visibility beyond keywords and clicks. It explores what this shift means for data, brands, and the future role of SEOs.
Here are the core themes we explore:
Google vs. ChatGPT in AI search: evaluating market leadership and monetization as user behavior expands beyond classic search visibility.
Proprietary data in AI chat: reframing advertising viability through data access and exploring what analytics might become beyond keywords in AI chats
Entity-driven SEO in the age of AI: letting go of keyword-based mental models as topics, intent, and structured brand signals matter more in AI-first environments.
The paradox of SEO's evolution: SEO becomes more verticalized while also expanding into brand, PR, and social influence.
The rise of agentic search: a long-term shift toward AI-driven transactions, requiring technical readiness, evolving metrics such as agentic completion rate, partnerships, and user trust.
Kevin’s most accurate predictions: AI’s real-world impact across industries, with a parallel shift toward workflow automation as clicks decline and SEO moves from clicks to influence.
AI-driven evolution of SEO: reframing the role of SEOs from strategic advisors to hands-on operators who leverage AI to prototype content, build internal tools, and directly influence execution.
Brand content as AI strategy: surfacing internal docs to boost discoverability, train LLMs, and defend reputation through context engineering and trust mechanisms like critical agents.
Time flew by in this conversation, and I’m sure you’ll listen with interest.
Video Chapters
Transcript
Gianluca Fiorelli: Hi, I’m Gianluca Fiorelli, and welcome back to The Search Session. Today, we have a guest who many of you would probably agree is the perfect fit for this podcast. And actually, you’re going to recognize him very, very soon.
He’s worked at G2, and he’s worked at Shopify. Right now, he’s a growth advisor for some very prestigious brands—like Snapchat, G2 again, Ironclad, AirOps, and many others. He’s also a Hypergrowth Partner. And this person is Kevin Indig, finally on The Search Session! How are you doing?
Kevin Indig: Hello! It’s good to be here, finally.
Gianluca Fiorelli: So, how are things treating you? How’s SEO treating you lately?
Kevin Indig: SEO’s been treating me well. I mean, you know, it’s a little bit like a marriage that’s been around for a while, and lately it’s gotten some fresh wind. A lot’s been happening lately, so it’s good to be in this game right now.
Gianluca Fiorelli: Yes, and it’s already been almost a year. In February, it’ll be one year since the eruption of AI into everyday life. I mean, it’s not like it wasn’t present before, but now, with AI Overviews in Search, formally announced, etc., etc., it really became, “Whoa, now this is real.”
And starting from there, it’s truly been a rollercoaster of news, ideas, and at times, even disillusionment with what was coming out. But surely, for us, and even more so for businesses, it’s been a very, very surprising time. Sometimes confusing, but also, probably, more exciting than the last few years, when we were sort of stuck in the commoditized, classic version of SEO.
Kevin Indig: Much more. Much more. Yes, it’s been a wild rollercoaster. And Gianluca, I have a feeling the next year is going to be similar to that.
Google vs ChatGPT: Who Wins AI Search?
Gianluca Fiorelli: And I think it’s going to be even more. Not necessarily from the industry side of AI players, because obviously, the fight between Gemini and ChatGPT is going to get even more vibrant, in my opinion.
I'm not sure if you agree with me, but Claude is very niche. And it seems like it’s also choosing to go very niche, and probably that’s the right decision for a company like Anthropic. It helps them maintain their uniqueness and position themselves more as a tool rather than something that everyone uses.
And then there’s Perplexity. It has many good things there, but also, in my opinion, many flaws. So I don’t know if Perplexity is going to resist.
So let’s go straight to it: which one will win, Google or ChatGPT? I have a strong opinion that Google is going to be the ultimate winner. It’s just too big to fail. And let’s say, after the Bard experiment, and then the first Gemini experience, even if we consider how AI Overviews started, and compare that to what they are now, with the experimental Web Guide, and the AI Mode improving almost daily, I think ChatGPT has fewer weapons to fight with, compared to Google.
So, I see the transition from classic search to AI search as Google being Google and ChatGPT as the Bing of AI search. Better than Bing in terms of volume and market share. Not by a huge margin, but still, a good second one.
Kevin Indig: ChatGPT obviously has a very close partnership with Bing, so it would make sense for them to also be the perceived Bing of AI search.
But look, it’s interesting, because last year, I wasn’t that bullish on Google, because, you know, it’s very easy for incumbents to lose the game, right? There’s this interesting paradox: people often say, “Oh, you should bet on the underdog,” because obviously, the underdog is in a weaker position. But I think it’s actually more interesting to bet on the incumbent—or the winner—because they can only lose. They can’t gain more. So, in this situation, it’s Google’s game to lose.
And as you said, in those first couple of years, they were really caught off guard. They weren’t executing well. But that has very much changed—I would say, especially in 2025, and particularly over the last couple of months with the release of Gemini 3, which is absolutely amazing. And they still have some more to go.
So I agree, it’s Google’s game to lose. I think they’re in the absolute strongest position from all the different layers that touch on LLMs: from a hardware perspective, a data availability perspective, a mindshare perspective, usage, feedback loops, and reinforcement learning, right? All of these things that matter, Google is in the best position to own.
And yet, ChatGPT has over 800 million weekly active users. So what ChatGPT has is momentum. It’s seen as the cool new kid on the block, which can make a big difference, right? It’s surprising how many regular people—in my surroundings, in my closer network of family, friends, and others—actually use ChatGPT. It’s absolutely incredible.
And so, as a third point, I think ChatGPT definitely has more of that subscription readiness from a monetization perspective. People are more willing to pay for ChatGPT.
Now, it’s interesting because Anthropic, as you said, is focused on enterprise and work. And they’re super strong there, right? They’re going public next year. I would love to invest. ChatGPT has that kind of new coolness factor. They’re growing rapidly, and I think their biggest challenge is to introduce an advertising model. I think that would make things more interesting and improve the overall experience.
And then, on the other hand, Google needs to figure out AI monetization—which I think they will do—but they almost need to figure out how to sell more subscriptions, right?
So each of these players is kind of in an interesting position. Let’s see how it shakes out.
The data shows that Google and ChatGPT don’t seem to cannibalize each other, right? It seems like people who use ChatGPT also use Google. And it doesn’t seem like there has been a decrease in Google usage—which, I mean, you could argue a little bit.
But if that’s true, it’s interesting because it would mean the pie is expanding for SEO. It means that we have more surfaces to play on. And it increases the overall value of what we do. So yes, it’s an interesting time, man.
Gianluca Fiorelli: Yes, totally. Returning to the race for AI domination. Obviously, it’s not a perfect comparison, but it reminds me of the race for mobile domination.
Apple had great design, a great interface, and came first. It could have been considered the cool kid on the block. And Google, in that case, didn’t create something new but rather bought something, which was Android. Later came the first Android version and the first Android phones, which, sincerely, were ugly and not easy to use. But then, they were able to create collaborations with great companies—like Samsung, for instance—and now Android is dominating.
But there’s another thing—if you remember—about Android. Android is the operating system, not just for mobile phones. It’s on televisions; it’s in many, many things. And I think that, if we don’t think just in terms of SEO and search, Google is trying to replicate the same thing with Gemini.
Apart from the classic Gemini that lives in every Google application, there are specific models for health, for universities, for almost everything. So, monetizing that kind of super diversification—creating specific models for each type of industry, which can be used professionally—maybe that’s something it doesn’t seem to me that ChatGPT is trying to do. That kind of evolution.
So ChatGPT is going for the big public. And what you were saying—yes, now when people think of artificial intelligence and search, it’s directly associated with ChatGPT, which is a cultural win, if we think. And maybe that’s the biggest challenge for Google.
And coming to the advertisement situation, the riddle is, how do you put advertising in a way that is clearly recognized as advertising? Because of legislation, of course, to avoid legal issues, and also to avoid deceiving users. But also, how and where do you put it? Because it’s honestly difficult—especially in terms of textual use of a chunk. Let’s say you are taking a chunk from a landing page that is paid, and you have to display it. Maybe they can invent some sort of new paragraph at the end, something like, “This recommendation is from our commercial partners.”
So, something like this could be an idea. For Google, it is easier in other cases—especially for shopping—because they can feed Gemini with the shopping feed. That’s why shopping was the first thing to come out in Google’s AI Mode.
What Analytics Could Look Like in AI Search
Gianluca Fiorelli: And do you think that, whenever they finally figure out how to put ads in the chat box, we’ll finally see some sort of proprietary analytics from them? Because that’s the biggest problem: how can you sell something as an advertising space without giving any information about how your advertisement is working?
Kevin Indig: You have to. So yes, I think that will be the key for us to get more proprietary data. And the very interesting question is, what will that data look like?
Because if you think about it, it’s easy to aggregate classic search keywords—because they’re short, right? If you have, say, three to four, maybe five terms, it’s going to be searched by thousands or millions of people, depending on what we’re looking for.
Now, prompts are different. Prompts are, on average, five times longer—sometimes way more than that. And that means there’s less uniformity between prompts. So it’s really hard if you have, say, a 30-term prompt—which is not uncommon—it’s very hard to say, “Oh yes, you know, 20 people looked for that, so that’s a topic you should go after.” Like, it doesn’t make sense. Prompts are too noisy. There’s too much variability.
And so, what I think makes much more sense from a data perspective—even for advertising—is topic signals and intent signals, right? So, almost like Google Trends. Here are the topics that people have searched for over time, and here is their intent.
Gianluca Fiorelli: Oh.
Kevin Indig: And so, bringing it back to advertising, I think starting with ads for the highest purchase-intent queries—or prompts and topics—makes the most sense, right? So, similar to Google, I think ads will start with these high-intent prompts, or topics—better said—that will be aggregated into topics.
And that will be highly interesting, because it will reveal how much intent power there really is behind ChatGPT, right?
We’ve probably all seen the research papers about how people search—or use—ChatGPT. And it’s interesting, because you can see people use it for all sorts of things. It’s not a search engine, right? It gives advice, and it helps you do certain things. It’s a little bit like an agent sometimes, but it can also be a tutorial; it can help you find certain things, it can help you analyze pictures, etc., etc.
And so the big question is how much of all this usage on ChatGPT—which must be massive—is actually purchase intent? We’ll find out soon, because I’m sure advertising will come, probably next year. And I think the data will be split into topics and intents.
Gianluca Fiorelli: Yes, I think the same. Somehow, very few commercial tools that exist right now are really treating AI search in terms of topics. Most tools are still prompt-oriented, which I don’t really like.
I mean, I’m really skeptical, like you. And I think that beyond just topics and intent—because the other big challenge is personalization—maybe they’ll eventually give us super-anonymized demographic data, in order to understand things like this topic + this intent + this kind of demographic = it’s working well, versus same topic + same intent + different demographic = not working well. That would be super useful for us in SEO, too—so we can better fine-tune our strategy, the content strategy.
Gianluca Fiorelli: And when you were saying “like Google Trends,”—I mean, I tend to get into a kind of conspiranoia—which is the combination of conspiracy theory and paranoia—I wondered why they waited more than 10 years to give us the API for Google Trends? Maybe it’s because they themselves finally needed an API for Google Trends. That could be an interesting little conspiracy theory to investigate.
Kevin Indig: Yes. Well, that they also aggregate more search terms in Search Console into topics.
Gianluca Fiorelli: Yes, in fact. I think—and I find it very weird—that the SEO community didn’t spend much time talking about it. Because it apparently seems like a simple filter—for instance, a brand filter for web brand queries.
But if you go read more carefully, it’s not just the classic filter that we can do manually with the RegEx. It’s really entity recognition. So not only your exact brand name, and not only all the misspellings, but also—for example, in e-commerce—all the products that Google considers part of that brand.
Kevin Indig: Yes.
Gianluca Fiorelli: Like, think about Nike. Google includes all the product names—specific sneakers like Air Jordan, etc., etc., etc. Which is interesting, because it’s clearly telling us that entity search is entity search. Now we don’t have to doubt data anymore. And secondly, that this is probably also used to understand whether a brand is a real brand or not.
Kevin Indig: Yes, a hundred percent. I think it sends some interesting implicit signals. I hope we can get that type of data through the API as well, because Search Console is heavily sampled, and the user interface is very difficult to use.
But to your point, Gianluca, I think it’s pretty clear that we’re going across classic Google Search and AI Search into this direction of topics, intent, and entities.
And I mean, you could argue we’ve been there for a long time already, right? But it’s still interesting that we’re so attached to classic SEO mental models in so many ways. It’s not just keywords—but also this idea of search terms, ranking factors, and all that kind of stuff.
And we need to shut this off, right? We need to evolve. We need to adapt to new ways of thinking. And it’s difficult, because I catch myself all the time having the reflex to solve an AI search problem using classic SEO thinking. And it’s like, “Oh no, wait. I have to take a step back here. I can’t use the same tools, because this stuff is evolving.”
I don’t want to say it’s completely different, but the way I think about SEO versus GEO or AIO—or whatever you want to call it—is that it's pretty much the same tactics, but in different environments and on a different playing field.
Gianluca Fiorelli: Yes. I mean, I don’t want to enter into a sterile polemic about the terms. And yes, it’s a new surface, as you were saying before. It’s like when the News Search appeared, it was a new interface, with a different algorithm and different rules to follow.
The biggest example of that now is the big difference between classic search and… Ironically, did you notice the experiment in Google Search with Web Guide?
Kevin Indig: Yes.
Gianluca Fiorelli: And did you notice, above the block of the Web Guide, it says “Return to classic search”?
Kevin Indig: Right.
Gianluca Fiorelli: This is fun, because it’s how we think in classic search. But I was saying that even with all these things, I still think local search and classic search are the biggest examples of similarity between classic search and AI search. So it’s not that different.
But then, you have people really specializing in local search—just like you might have, maybe in the future, or especially now, specialists in AI search. Because maybe in the future, classic search won’t even exist anymore.
I mean, I’ve always thought of SEO as something bigger than just the classic search we see in the old view. So that’s why all this debate and polemic over terminology, for me, is like staring at your finger instead of looking at the whole picture. It’s so stupid.
Kevin Indig: Yes. It’s a semantic philosophical conversation. But you’re making this a really interesting point—which is that we’re seeing this paradox between the verticalization of SEO, right? So for a while now, as you said, we have news, we have shopping, we have local, you know, B2B, B2C. They all follow similar principles, but they’re different games to play.
Gianluca Fiorelli: Yes.
Kevin Indig: That’s one side. But then, on the other hand, at the same time, SEOs—as a craft, as a role—we need to do more and more work in other departments: brand, PR, and social media, right? More than ever before. And so you have that weird paradox: it’s getting more verticalized but also broader at the same time.
Gianluca Fiorelli: Yes, totally agree. But I think there’s also a risk—which is innate, I think, in the nature of SEOs—to try to appropriate the leadership in everything. When in this case, it’s not just SEO leading. It should also be other marketers from other fields. Because otherwise, it’s going to lead to a clash, an immediate clash.
It’s doing something that maybe a few SEOs have done, but I think the majority never really did: collaborate with the other channels, to retrofit the work across all the growth strategies between all the channels—in order to do the classic one plus one is not two, it’s three. And in this case, it could even be four or five.
Kevin Indig: Right.
Gianluca Fiorelli: And this is the most important thing. So, for instance, I would never have really thought to include, in my workflow when I’m creating a strategy, a chapter dedicated to amplification across other channels, so the classic omnichannel.
I would’ve said, “Okay, when you’re creating this kind of content, it would be better to start thinking about how to repurpose it for all these platforms.” Just that.
But now, I’m also trying to say, “Okay, the best way to approach this—consider that, for instance, Twitter is present in Google Discover, and that LinkedIn, Reddit, and Quora are super used by AI as sources. Let me tell you what kind of style you should use, which works better than others.” And so this is the new part. Especially.
Agentic Search: Optimization and Hurdles
Gianluca Fiorelli: And one other thing. At the very beginning, I said to you, I think 2026 is going to be even more frenzied. And the reason is—especially in the e-commerce space—maybe not in the first half of the year. For the first half of the year, I think and foresee big brands, especially, really working on setting the base for agentic search.
Because the problem, especially, is technical—to make a website, especially an e-commerce one, comply with all the needs that are required by the agentic search.
But for the second half of the year, I see an explosion of everything agentic. Because there will be a sort of convergence between the experimentation ChatGPT is doing, the experimentation Google is doing—also in AI Mode and eventually with Gemini—and how the big brands, especially the leading ones, along with some good, let’s say, lone wolf smaller brands, want to experiment with this too. And maybe they can do so with fewer constraints—less bureaucracy—than the big brands have to deal with.
And I think that will be the big topic of frenzy in terms of AI and search. Because in that case, yes, we will need to return to this verticalization and specialization in SEO. We will need someone who is really going to specialize in agentic research. Because it’s not the same as SEO as we’ve done it before. That’s a big difference, more than AI search now, in fact.
Kevin Indig: Yes. Look, I actually thought we were going to have this wave of agent usage already this year. It was too early. These agents are not good enough yet. You know—like Google Mariner, and the ChatGPT agents—they’re still crashing a lot of the time. They’re getting better, though. And so we can see that this is coming on the horizon. It’s coming down the pipe.
The question is, when are we going to adopt something like agentic completion rate as a metric? This is something I think we’re going to talk and think about a lot more next year. What’s the rate of agentic visits to my site compared to how many agents finish a checkout or complete a task? That’s going to matter.
And we’re going to notice, along the way, that there are some technical hurdles these agents have to overcome. There are things like popups, or other navigation elements, maybe JavaScript-heavy navigation, that are going to be difficult for agents to handle. Cookie banners, right? These classic types of things are going to make agents crash, at least in the early stages.
But also stuff like slow server response time or certain Core Web Vital issues. These things are going to matter more for agents, because they’re kind of like a newborn giraffe, right? They’re always at risk of falling over and crashing at any second. So you really need to make it very simple for them.
Which also brings up questions about semantic HTML and all that kind of stuff. Sure. But also CSS usage and JavaScript usage. How interactive is the site? How easy is it to navigate? All these kinds of things.
On the other hand, I also think we’re going to talk more about agentic schema, not just in the classic sense of schema markup. But there might be new ways to create paths for agents with specific schema. You have that already, for example, with Siri or Alexa. There’s a specific voice assistant schema that allows them to take certain actions.
And so I think that’s all coming down the pipe. I agree with you, I think this is more likely to happen in the latter half of next year. But it makes a lot of sense to go in that direction. On one hand, to increase the utility of AI. It makes it much more useful if the agent can do something for you immediately.
And also, on the other hand, when you think about model training, we went from this progression of basic LLMs to reasoning models, more reinforcement learning and as the next step, there’s going to be more agentic learning and agentic development on top of LLMs, just to increase their utility.
Because companies are putting a lot of money into this stuff, and it better, you know, deliver some economic returns.
Gianluca Fiorelli: Totally. And there’s another aspect that maybe is going to be very steady—and worth analyzing—which is, how do you make people accept it?
I mean normal people, like me and you, when we’re not in the marketer mindset. When we just want to buy something for our kids, for people we love, or for ourselves.
So how do we educate people to decide and accept the idea that they’re not doing the task themselves but delegating it to something else? Because obviously, someone could say, “But people are already used to this kind of thing, somehow.” Like when they shop on Amazon. Or when they use Google Shopping.
But anyway, at the end, people always—if they haven’t stored their credit card number in Chrome—usually end up putting in the credit card manually. So, it’s this kind of control. Maybe it’s more about how to create trust and how to create usage habits.
So I think, in terms of big industry movements, we’re going to see something that is already happening, like the Stripe deal with—was it ChatGPT or Perplexity? I don’t remember now. I think it was ChatGPT.
Kevin Indig: ChatGPT, yes.
Gianluca Fiorelli: But maybe also on a more local market level, with big banks. For instance, here in Spain, we have La Caixa, and we have BBVA. They are really, really strong on the internet banking side. So maybe we’ll see them creating big collaborations with AI agents—AI players—in order to build this ecosystem that’s also convenient for the financial industry.
So I think we’ll see those kinds of developments too. Because, I mean, we can be fully prepared for agents, we can do everything cool with agentic search, but if people don’t trust agents enough to give them everything, then it’s going to die.
Kevin Indig: Yes, sure. Trust is everything, right? Even in search—if Google loses trust with searchers in the search results, if people stop believing these are really the best results—that’s probably the worst thing that can happen to Google. One of the worst things.
And so you're absolutely right: trust is at the core of this. And trust is built over time. That’s why, the better these agents become, the more willing we’ll be to give them our credit cards. And we’ll limit it in the beginning, right?
It’s like when you have a corporate credit card, you have free spending up to 10 euros or whatever, and after that, it asks for permission. And then, over time, we increase the spending limits.
And yes, there will be cases where people lose a lot of money because they’re not careful, or they make mistakes. Yes. But over time—and I don’t know how long that takes—it could take, I don’t know; it’s hard to say because there’s so much nuance. But I think fully fledged mass adoption will probably take a decade or so. People move slower than technology.
Gianluca Fiorelli: I mean, we needed a pandemic for people to fully go into and trust e-commerce.
Kevin Indig: There's still a way to go, right? But yes, you're absolutely right. So, you know, there are these catalyst moments, and there are these slow developments, and we'll eventually land there.
So on the one hand, I think this is something we should start thinking about—agent e-commerce. And at the same time, I also don't want to join the fearmongering and say, “Oh, you need to do this tomorrow; otherwise, your company’s going to die.” It's going to take a lot of time, right? So, start thinking about this, but don’t go to bed worrying about it. We’ll transition.
Gianluca Fiorelli: Obviously, it would be foolish and crazy if, let’s say, a big e-commerce—or even a small e-commerce—was going full agent and totally forgetting that still the biggest amount of traffic and conversions comes from classic search.
And as I like to define it, The Search Session is a place where you start thinking forward in order to study and get prepared for things that are going to come. And eventually, the people listening to us will have a competitive advantage.
As I said before, that doesn’t mean you have to stop doing classic SEO. I mean, technical SEO is even more important than before. And we’re just saying that it’s going to be even more important in the future. And doing things like great content, organized architecture, internal linking—those are still, still, still important.
And I personally think they are going to be very important also in the future—for different reasons, but still important. So yes. And, talking about the preview you did for 2025. You just said you were thinking that agentic search was going to be a bigger movement in 2025. But what is the one you are particularly proud of? The one you guessed well.
Kevin Indig: Man, what did I guess well? Let me think a little bit about this. I think, you know, there's something to be said about the fact that we saw more "victims" of AI. And I’m not proud of the fact that there are more AI victims. I guess I’m more proud of the fact that I held onto that prediction, or that outcome.
But I think the reason that's interesting is because, on the one hand, we're saying, "Look, AI takes time. Classic SEO is still important, or classic search results are still important.”
And at the same time, there are real industries and real companies where we already see a negative impact from AI. Right? Like translation services, Stack Overflow and Chegg are clearly suffering, right? So I think that’s one prediction that has come true. It was kind of foreseeable but also has this logical unity—you either believe that there’s a real impact of AI, or you don’t. So that one landed well, I think.
Workflow Automation: From Clicks to Influence
Kevin Indig: And then the other one is that more marketing teams are using workflow automation. I always say that AI goes both ways, right? It changes the search landscape, but it also changes the way we work. Something that really came to light this year is that so many more teams are not just using AI but are using it to automate parts of their work.
And I think that’s important, because with the decline of clicks, we’re losing an important leading indicator. So we need to be even more efficient with our work in order to be successful during this transition. Because all the leaders are now wondering, “Okay, traffic is down—what’s going wrong?” And we need to help them understand that this is changing from a click game to an influence game. And automation helps with that.
Gianluca Fiorelli: Yes. And regarding using AI for work—especially in our case, as agencies or even consultants like me—I’ve also started to notice myself using more AI in my workflows. But I’ve also seen a sort of returning wave. I mean, at the very beginning it was like, “Okay, you can do this, this, and this—almost automate everything.” But now I see a shift back—like, “Yes, you can do all these things, but we need to put the human in the loop.”
Kevin Indig: Yes, sure.
Gianluca Fiorelli: Which maybe was something that we should always have taken for granted—I mean, as like a default. There should always be a human in the loop. Even in plants with robots building cars, you always have a human in the loop, controlling that everything is fine.
But now it’s coming back. And it’s coming back, especially on the content side of using AI, because people have already had time to see both success and immediate failure when creating content with AI.
And this is one thing where certain kinds of professionals were really hit by the eruption of ChatGPT, especially two years ago, for instance—they are maybe going to have a new opportunity to be supervisors. Not necessarily as the person writing the content, but as the person guiding AI on how to write things.
Not in terms of a “prompt engineer,” but more in the sense of—like, you know—an editor-in-chief in a newspaper.
Kevin Indig: Yes, I love that comparison. That’s a brilliant analogy. It's really what you're talking about. You know, the more we progress with AI, the more I’m convinced that humans will stay an integral part of this work. And they will become so much more efficient.
Now, the big question is obviously how much will AI replace us—jobs, people? And I’m more and more convinced that yes, it’s absolutely going to replace some jobs, but I don’t think it will result in this massive replacement, though.
I think what's going to happen instead is this: first of all, there are going to be some people who will be 10 times, maybe 100 times, more effective than others—because they're better at using AI.
But second, we’re all going to become more like directors of AI and conductors, or editors-in-chief. And we're just going to increase our output, right? And I’m saying that because there used to be this interesting theory—I think in the early 1900s—where it was clear that technology was evolving. And so the theory then was that humans are just going to work three to four hours every day. What has happened instead is that humans are working more than ever before.
Gianluca Fiorelli: Yes.
Kevin Indig: I think that—to me—is kind of a little bit of an indicator, maybe even a proof point, that with AI, I don’t think we’re going to keep our productivity level or productivity growth the same and then employ fewer people. I think we’re going to employ the same amount of people, but our productivity is going to go through the roof. So anyway, that’s my TED Talk.
Gianluca Fiorelli: Yes, if I look at myself, for instance—taking me as a very improbable, unstatistical case—I’ve never worked so many hours. Because now I can do more things.
So I can dedicate time to more things—not just because I’ve quit the more repetitive tasks that are now successfully automated—but because AI is giving me tools that I wasn’t able to access before. Or that, before, were costing me too much time. So I used to dedicate that kind of work only for very specific cases.
Now I can, let’s say, widen my landscape of work. For instance, I’m really into things like content clustering—very, very specific content clustering. Before, maybe I would just say, “Okay, let’s do content clustering for the classic topical content clustering. And maybe let’s try to do the clustering expanding to a search journey, a customer journey, not just targeting informational content.”
But now, there is the possibility to inject into AI very, very specific data about buyer personas and psychographic data. So really, by using tools like SerpAPI.com, scraping the SERPs at scale, I can understand what features tend to appear the most. So let’s say for this kind of search, for this kind of micro-moment, is Google usually showing videos? Is it showing images? Or is it showing something else?
All this data I can now collect much faster. And so I can inform the AI—and with me, the human in the loop—to try to build stronger clustering. Doing more clustering, and eventually—which is something I’m a bit crazy about and starting to experiment with—how to cluster the clusters.
Because otherwise, you’re just going to end up with silos. So, how can you connect the silos if they can be related?
These are things that I always had in mind, but before, they were painfully manual to do. Now I can delegate many, many tasks to Claude, to Gemini, or less to ChatGPT and more to Claude and Gemini in this case.
So yes, I’m doing more things—and sometimes working more hours—because I’m really investing time into this kind of work.
Kevin Indig: Totally, and once that workflow is built and refined, you can use it over and over again, right? So you're building more internal tools, and you're exploring new concepts.
And what I’m really excited about personally is that SEO becomes less of a set of recommendations or principles and becomes more of a discipline where SEOs can also have an output.
Because for the last two decades or more, SEOs gave recommendations. Now, they can actually create content based on clusters, all these cool analyses and workflows.
And now we can actually bring the content all the way to the team—maybe even theoretically publish it—or, even better, bring it to the editor and say, “Look, this is a very refined draft. I’d love you to make it better.” Or at least, provide a very, very detailed content brief.
So we can start to have real influence. The same with CMS, you can plug more AI tooling or workflow automation directly into the CMS. You can now create your own components just with natural language, right? There are things like Figma Make or some of the Photoshop tools, where you can design something and then code it with Cursor and bring it all the way to a company.
And again, I’m not saying it should be immediately pushed live; the designer should improve it.
Gianluca Fiorelli: No, no, it’s an easy way to finally create prototypes that can sell your recommendation better.
Kevin Indig: Yes, exactly. I can spin up a website with Lovable in 10 minutes, right? Like, if I have a knowledge base attached to it and some stylistic guidelines, I can create programmatic pages in no time. And again, all at the prototype stage—but that is now possible as an SEO.
We're not as reliant on other teams anymore. We’re still—look—we're not bypassing them, right? We still need to collaborate on everything. But we can now move much, much faster. And that’s huge for our profession.
Context Engineering: Unlocking Internal Data
Gianluca Fiorelli: Yes, totally. And one question, now that you are consulting as an advisor. This is really a personal curiosity, because it’s happening to me too. But how many times did you find yourself pointing with your finger to your clients, saying, “Hey, you don’t need to create great content. You already have it! But you have it already on these subdomains, in these PDFs, in these things that are totally on your website but totally orphaned.” How many times has that happened to you?
Kevin Indig: Every time because you're absolutely spot on. It's not marketing collateral and sales collateral. It's also help documentation, internal documentation, right?
Gianluca Fiorelli: Exactly.
Kevin Indig: There's so much content. For example, you could plug into Confluence, right? Atlassian. I worked there. It's used a lot of times by companies as, like, an internal knowledge base. The content ideas and content you can create just based on that internal stuff, right? I’m not talking about sensitive data.
Gianluca Fiorelli: Obviously.
Kevin Indig: Just the context. It's massive. And it will take companies years to bring that treasure chest from the ground of the ocean to the top. But you're absolutely right. That's why I like the term "context engineering," because so much context already exists that is not yet explicitly formalized in external-facing content.
Reputation Management and "Reliability Racks"
Gianluca Fiorelli: Yes. And I think it's also a good way to plan defensively—in terms of a defensive plan—because I think you agree with me that LLMs, the chatbots are, for brands, essentially a very sophisticated online reputation management surface.
And if you have everything to substantially make the opinions of others about something—they are going to be surfaced, obviously—but maybe, if you're making surface in, for instance, all your customer care help subdomain, your developer subdomain, and your sections that talk about your brand—obviously, as you said, without putting in sensitive data—but everything that can really explain your brand to AI, then AI is going to use your facts instead of using facts from others.
Or maybe worse, facts made up by others—that’s the real problem. It’s something we’re already seeing with tests, for instance, from Ahrefs, or that kind of situation.
Kevin Indig: Yes, totally. I mean, it's a real problem right now. As you said, Ahrefs has published an article about a manufactured imaginary company and fed LLMs information. I've noticed several instances in which a company’s reputation was severely damaged from external sources of the company.
So all of this is real, and that is one of the reasons why I think that starting next year, more LLMs are going to adopt, you know, agent debate models or critical agent models. There's also a concept called reliability rack, which all mean the same thing—basically, instead of just having LLMs retrieve content for training data and for reasoning answers, these systems will add filters—little agents that reason about the content being retrieved. So they’re critically asking, “Hey, where is this content coming from? Is there some prompt injection in here?”
Gianluca Fiorelli: A sort of grader.
Kevin Indig: Yes, exactly. Exactly, it’s going to be a grader. I think Amazon AWS already provides this kind of technology called Reliability Rack. And then there are a bunch of research papers about this idea of critical agents or agent-to-agent debates. Right?
And so, because of all these issues—also with reputation and whatnot—I think LLMs are going to become more sophisticated, especially when it comes to web retrieval or web search.
Gianluca Fiorelli: Yes. I personally also think that something that is not new—actually, one of the oldest technologies—which is symbolic AI, is going to be recovered. Because it runs substantially on knowledge graphs. And maybe knowledge graphs can be one of the ways for AI agents to understand if something is authoritative or not. Or whether it comes from an authoritative source or not.
There are many other things that LLMs must also recheck. Like this—for me, it’s really annoying—the over-dependency on recency. Because it’s not possible that—let’s say, I’m making up an absolute example—the Encyclopedia Britannica is considered less relevant than a post by a blogger, just because the blogger’s post was written two days ago, and Britannica has always been there.
So, there are also other kinds of problems like this one, which should be reviewed. Because, you see, the recency is also prone to spam. You can automate bots to create listicle spam, and voilà—some non-existing brand ends up being suggested on the same level as a historical, reputable brand.
Visual Search, Smart Glasses, and Google Lens
Gianluca Fiorelli: Just one question. Earlier, we were talking about omnichannel—about how to collaborate with other digital channels. But I foresee an even stronger collaboration with classic marketing strategy—I mean the offline side, let’s call it. Where we can see that what’s online, the brand visibility online, is going to be a reflection, a true reflection, of what the real visibility offline is.
And I say this also for the potential evolution of the devices themselves. So maybe—we used to think of voice as always kind of a joke—but now we should really start to study how many people are actually using ChatGPT, for instance, as a voice search device instead of a typing device.
Because, I mean—who can humanly write a 20-minute-long conversation by typing? Nobody. And people forget that for over 10 years now—especially the younger generation—people have been using WhatsApp voice messages, which I hate! But it’s something that’s widely used.
And then, maybe also because we’re starting to see things—we haven’t even talked about Meta—and Meta is pushing all this technology around, let’s call it, glasses. But it could also be AR on mobile, or something like that.
So we haven’t talked about Lens and visual search either—as search, even as a more improved surface for search. So I think this is something that not so many people are talking or thinking about, but it would really be worth investigating.
Kevin Indig: Let’s see, next year, Google is going to launch glasses—AR glasses—so that’s going to be highly interesting. I expect Apple, if they can work through their leadership challenges, to release a more powerful Siri that hopefully gets really good. And at the same time, Google has been publicly announcing that they’re seeing, I think, billions of searches through Lens.
Gianluca Fiorelli: Yes.
Kevin Indig: And I wonder, what are these people searching for, right? Like, is this really just like, “What’s this tree? What’s this plant?” or is it something else?
Gianluca Fiorelli: I have a passion for Lens. I was talking about Lens back in, let’s say, 2018—already when it was just coming out—and it’s huge for commerce, especially for showrooming.
Like when you go to a shop and say, “Okay, this is something I like. But let’s see if I can buy it online”—maybe from the same company, from the same shop online, or if there are offers for this product from other eCommerce platforms. That’s quite common.
And then, you know, the fact that Lens is also a prototype of what could be AI search before its time. Because people were already using Lens. For example, you can use it when you see something written—let’s say you’re in Japan and you don’t know Japanese—and with Lens, you can simply take a photo and ask for a translation.
Or, which is another cool thing you can do with Lens, and they know that people are using it for that. For example, taking a photo of—let’s say—a restaurant. And instead of going to Google Maps and searching for the restaurant, they’re asking directly, “What restaurant is this, and what are the reviews?” So it’s really simplifying certain types of search. That’s why it’s so, so big, especially for the younger generation.
Kevin Indig: Yes. So it’s interesting, because either we’re seeing substantial e-commerce usage for Lens, and then the question is, how does that translate into revenue for Google? Right? Or it’s more simplistic, informational types of searches.
So let’s say there’s a significant amount of these searches, of these billions of Lens searches, that are e-commerce related, even if it’s just 20%, right? It’s still a massive number. And so, Google probably sends them to a SERP for Shopping, and that’s where they can show ads.
And since we know Google’s revenue has been growing, right, but clicks are down—because of AI Overviews—the question then is, are these searches from Lens in e-commerce the reason revenue is still growing, on top of classic search? Or is classic search eroding, and Lens is compensating for some of that erosion and driving more e-commerce value? We probably don’t know the answer, right?
Gianluca Fiorelli: No.
Kevin Indig: But I’d love to understand what’s the impact of Lens search on Google’s bottom line? Or is it just a cool way to stay top of mind for users, and that’s the business value for Google?
Gianluca Fiorelli: Well. I sincerely don’t have the answer. But I know that when you do, for instance, an eCommerce search using Lens, it’s like when you’re using image search on desktop or mobile. There are several risks for a brand. That’s why you can start seeing, for instance, very sophisticated packaging being very well retrieved by image recognition algorithms.
Because if not, it’s going to be confused—and maybe a product of your competitor is going to be suggested instead of yours when it's been photographed. This is one problem.
And the other problem is that Google is not stupid. If you don’t like this product, they’ll present you with related products that you might like, want to see, or explore. So Google is always—you know, like in a casino, it is the bank. It always wins.
What is risky for the organic search side of the brand, and, as it is in Shopping for the paid side of the brand. Because it’s so simple in Shopping. You have your product or your product knowledge panel, and then Google is suggesting another shop, another brand. Why?
Beyond the Bio: Kevin Indig
Gianluca Fiorelli: Okay. So, one hour already.
Kevin Indig: Time flies by.
Gianluca Fiorelli: So let’s stop talking about the AI future and let’s talk about you. I always have this kind of question: what is the thing you would’ve liked to do if you weren’t a growth advisor? When you were a kid, and someone asked you, “Hey Kevin, what do you want to be when you grow up?”
Kevin Indig: You know, my answer most of the time was becoming a doctor, a surgeon. My dad is a doctor, and I wanted to become a doctor until I was about eighteen years old. And then I got really turned off by all the bureaucracy and paperwork you do as a doctor. You know, it’s probably 50% of the job now, and that didn’t seem so appealing. But you know, I think I would’ve had a great time as a doctor.
Then, right after college, I didn’t immediately get into SEO. I first started in finance—for like a year. Not long, but I was at an investment bank. And I thought that was the way to go, but then I was like, “Yes, no, that’s not—it’s the wrong kind of direction.”
You know, I was born and raised in Germany and then did investment banking in Switzerland for a year. But the action was not in Switzerland anymore, you know? The action is in the Bahamas, in Singapore, in Hong Kong, you know—other parts of the world. Maybe New York, of course.
So I then pivoted toward SEO, and I was very happy to have done that. But yes, you know, when kids asked me on the playground when I was young, I was like, “Yes, I’m going to become a doctor.”
Gianluca Fiorelli: Yes, well, like me, I wanted to be an archaeologist. Also because, I mean, I grew up when Indiana Jones was running in the cinema.
And I actually started to study, to do some kind of study, in order to be an archaeologist. But then I understood that—okay, Latin, somehow I can do it. But ancient Greek? Ancient Greek is not for me. And it’s mandatory to know ancient Greek if you want to be an archaeologist.
So I turned to my second passion, which was movies. And that’s why, for many years, I worked in television. And then, as I’ve said many times, I had to reinvent myself as an SEO, for reasons that it’s too long to explain here. And it's really not depending on me. It’s a really external reason. A really big business reason.
There is sometimes a confusion I have with you. You were born in Germany, so you are German. Because many times people think that you are not German. People think that you are from the States.
Kevin Indig: Yes, yes, that’s right. The thing is, my dad is American. So I’m a hybrid, if you will, German-American. I have dual citizenship. And yes, you’re right: I was born and raised in Germany. I spent the first 24 or 25 years of my life in Germany, and then I moved to the States.
I spent the last 12 years there, and now I came back—just this year—to Germany with my family to be closer to my parents and brother. So yes, it’s been, like, an interesting experience.
And then in the U.S., I spent a lot of time in California, a little bit in Chicago, and a little bit in Michigan. That’s why people are also confused. So I don’t make it easy for them.
Gianluca Fiorelli: No, sincerely not. I will give a suggestion for all the people that maybe want to contact Kevin: Kevin is in the Central European Time Zone, okay? Because sometimes you don’t know, “what time zone Kevin is in, when I can contact him, when I can organize a video call with him, and so on.”
Kevin Indig: It happens all the time. All the time, man. And I can’t blame people. Again, I don’t make it easy.
Gianluca Fiorelli: No! Well, it’s like for me, everybody thinks I’m Spanish, when instead I’m Italian. And they say, “Ah, but you’re Italian born in Spain?” “No, no, I’m Italian and moved to Spain.”
So Kevin, thank you so much for this conversation. It was really, really interesting, and I think the time flew because of that.
Kevin Indig: So fast, man! But I’m glad we finally could make it happen. I really enjoyed this as well. I could’ve probably gone on for another hour. I appreciate you having me on, man.
Gianluca Fiorelli: Thank you. And thanks to you guys, remember to ring the bell! You know, this is the classic moment I have to do: ring the bell in order to receive a notification of new episodes. And help me grow this channel—subscribe on YouTube, Spotify, and Apple Podcasts. 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.
stay in the loop