This is The Search Session, and I’m Gianluca Fiorelli. My guest is Myriam Jessier, an international SEO consultant and trainer currently based in Denmark.
Myriam explains how AI search understands and recommends brands, and why video has become one of the strongest formats LLMs can understand, cite, and use as evidence.
She also reveals what brands need to do to show up correctly, from clear signals and consistent information to quality visuals and enough depth for machines to trust them with confidence.
What you’ll get from this episode:
How search keeps repeating itself: AI search feels new, but it builds on older ways of organizing and understanding information that matter again as machines need clearer signals.
Why clear brand-owned content matters: when brands rely on “pure vibes and marketing adjectives” or lack official content, AI has less reliable information to extract and may invent answers or rely on inaccurate third-party sources.
How technical SEO becomes technical branding: technical SEOs have a new role in reducing hallucinations, protecting brand accuracy, managing bot access, and helping machines understand the brand correctly.
How visuals shape AI understanding: image quality, product details, emotional cues, and visual context help AI recognize brands, while poor imagery can lead to misrecognition, lost conversion, and weaker visibility.
How video becomes a source of truth for AI search: video combines audio, transcript, and visuals, giving LLMs richer signals than text alone, so brands need to front-load key details and show products clearly.
How familiarity and brand depth drive AI recommendations: LLMs use repeated mentions, accurate naming, and connected information to build consensus, understand what your brand owns, and avoid misattribution.
Why brands need to match the category before they stand out: AI needs to recognize the expected category signals first, then unique details can help the brand differentiate.
Listen in for a conversation that makes AI search a lot clearer.
Topics covered: AI search · technical SEO · technical branding · multimodal search · visual search · video optimization · LLM citations · brand-owned content · brand visibility · brand depth
About the Guest
Myriam Jessier
Search Engine Optimization Executive at Pragm
Myriam is an international SEO consultant and trainer with over 15 years of experience, specializing in technical marketing, AI search, and multimodal search.
Through her consulting and training work, Myriam helps teams strengthen their search performance, technical marketing skills, and understanding of how brands are interpreted by search and AI systems.
In 2016 she co-founded Pragm with Augustin Delporte, a boutique technical marketing agency focused on SEO, Google Ads, analytics, BigQuery, and training.
Myriam is also an international speaker and writer, regularly contributing to Search Engine Land on some of today’s most important search topics.
Video Chapters
Transcript
Full conversation between Gianluca Fiorelli and Myriam Jessier.
Gianluca Fiorelli: Hi, I'm Gianluca Fiorelli, and welcome back to The Search Session. Today we are going to have—how can I introduce her? A guest who, like me, is not an English-speaking native. She has a wonderful, cool accent, a French one. She lived many, many years in Quebec, Montreal, Canada. She just told me that she relocated to Denmark.
You surely know her because she has a very strong personality, and she's very alive when you know her and you talk with her or when you see her speaking in the many conferences that she spoke at, like BrightonSEO, SMX, Voxxed Days Luxembourg, and she's also a contributor to Search Engine Land, with really cool articles.
She was one, or she is still one, of the few who pay due attention to multimodal search, and this person is Myriam Jessier. Hi, Myriam. How are you doing?
Myriam Jessier: Bonjour, bonjour. Very happy to be here.
Gianluca Fiorelli: I'm very happy to have you as my guest. So, how are things going there in Denmark?
Myriam Jessier: As you said, I move a lot. Currently, I'm in Austria.
Gianluca Fiorelli: Oh, okay.
Myriam Jessier: And in a few days, I'm back in Denmark. But I will say that things have been very interesting on and offline.
How Search Keeps Repeating Itself
Gianluca Fiorelli: Oh, cool, cool. Nice to know. And talking about online and our work, how is SEO treating you lately?
Myriam Jessier: I will say that SEO is treating me like it's always been treating me. It's very weird, and it's something that I love. I'm never bored. But I will say, since both of us have been in the industry for quite a while, I feel very comfortable discussing the fact that everybody on LinkedIn and other assorted social media seems to be quite busy arguing about whether we call it SEO or GEO, and I have reached a point where I do not care. And let me explain to you why.
Gianluca Fiorelli: Okay.
Myriam Jessier: The way SEO has been treating me is that the real question is what's happening to the web underneath that shift. What actually keeps me up is the bigger picture, because every choice we make as specialists in our industry either feeds the dead internet theory, that spiral, or pushes against it. So my endgame is to leave the web a better place than I found it. And lately, this seems to be a pivotal moment to ensure that we do just that.
Gianluca Fiorelli: Indeed, indeed. I think that—literally translating an expression from Italian—the polemic between SEO, GEO, and AEO has become a level that is like talking about the sex of angels.
Myriam Jessier: Yes. Yes.
Gianluca Fiorelli: Yes, you know, like in the Middle Ages, in the ivory tower of Alexandria, the philosophers were discussing how many angels could stay on the tip of a pin. And okay, it can be interesting internally in our industry, but after 18 months of continuing to talk about this, I think it's useless.
Let's see where the market is going, and you can think of it as something different. It's just something beneath the umbrella of SEO, like any other vertical it is, at least in this moment, where classic and AI search coexist.
But apart from that, the more important thing is understanding how it works. What is worth doing? What are the things that we did that still matter? And what are the things that we know we should have done and didn't do in the past but that matter even more now, and we should do now? Because there are tons of things that people always deprioritize because they are not relevant in that moment for poor ranking systems.
But now we know that this is important. And I don't want to get into the semantics. If we just think about technical SEO, we always said that having a wonderful, super-mega-fast website was not a strong ranking signal because Google also said this is not a strong ranking signal for classic search.
So we were saying, okay, but for conversion, it's very important. The page experience is very important and blah, blah, blah. And now we know that AI models, if a website is not super fast in responding, immediately discard the potential source that could be ours.
And so page speed, the Core Web Vitals, is a “ranking factor” for AI search because if you are not very fast, you are not going to be used as a source or as a potential source. And if you are not used as a potential source, you are going to disappear.
Myriam Jessier: I have this point of view because I like to tell people that I'm 56K-modem-noises old. I remember floppy disks. I remember my first internet connection. Yes. So this positions me in the history of the internet in one sentence.
And before being an SEO, I used to be a librarian, so I find it really amusing that everyone is recommending table of contents to be there. And I'm like, "Hmm, y'all forgot about this," but I remember that. And I remember seeing people transition from a catalog where they had to break their brains to search for things and ask a human for it, to "Never mind, I can ask Google."
I even had younger folks come to the library—so that was a long time ago, a very long time ago now, more than 20 years—and ask me, "How do I even search for a book? There's no Control + F." And I was like, "There is a thing called a table of contents, and there's a thing called an index. Check the book."
Gianluca Fiorelli: And it depends on which kind of index.
Myriam Jessier: Exactly. And so, yes.
Gianluca Fiorelli: Yes, because we have the classic index, which is the table of contents. Substantially, you have the chapters and so on. And then we have the real index, which lists every important voice and where you can find it. The bolded numbers are where we especially talk about it, unlike the non-bolded numbers.
I studied bibliotheconomy too and all this kind of differentiation between how to read a real index. And yes, that's why, maybe because I also come from that kind of study where I also had to study bibliotheconomy; I didn't do that kind of work, obviously, after.
But that's why I'm also very picky about talking about, let's say, architecture or taxonomy, and what is the ontology of these things? Because it all comes from these kinds of things: how to classify knowledge. And a website is knowledge, knowledge about my brand, my client's brand, the things it does, the things it doesn't do, the people working there, and so on and so on.
And so classifying and ordering things helps not only to work better—because that’s how I know exactly what I'm working on and trying to optimize for in my job—but also to present things through navigation in an easier way to the people visiting the website and the machine, now also an AI machine, to understand better how it conforms.
And not even talking about schema or structured data in the wider sense. I'm just talking about taxonomy and ontology. So, it is a pure simplification of the knowledge presented on the website.
Myriam Jessier: And to me, it's one of those very interesting things where I'm realizing that we're “discovering” new things that, to you and me, sound like concepts that we have already heard.
And I find it very important to say, to understand and make headway in an industry, the most underrated skill is to understand the history, understand the mythology, and understand the lore that brought us to this point. And if you invest the time to do this, it's very easy to see, "Huh, so this is where we're going. We've been through this before. This is an updated flavored version, but here we are."
So for me, for example, you and I are both image optimization nerds. And right now, when it comes to search, I find it amazing that it aligns more with human behavior. What I mean by this is normally, if you and I were seated in the same room and I saw something cool, I would go, "Hey, Gianluca, look at this." And your eyes would take in the movement, the images, the shapes, the everything, and I wouldn't necessarily need to say more than, "Look at this."
And now I can take my camera, my phone, open the fridge, show a video, and go, "Hey, go check the supermarket specials and create a weekly vegan menu based off of what you see in my fridge and what's on sale."
Gianluca Fiorelli: My case would be crazy because I try to invent something from an empty fridge.
As multimodal search and LLMs increasingly rely on video content, measuring visibility requires more than tracking traditional SERP features.
Advanced Web Ranking helps you monitor video performance across Google, Perplexity, and ChatGPT, track video features, and measure the impact of video citations on your cross-platform visibility.
Start your free Advanced Web Ranking trial to track video visibility across search engines and AI platforms.
From Technical SEO to "Technical Branding"
Myriam Jessier: Okay, hang on, because this is a very good joke, and it's very "apropos," as we say in French. Because when you tell me that the fridge is empty, this is how I feel when I visit some brand websites. I see their website, and I'm like, "An AI is supposed to provide answers based on pure vibes and marketing adjectives?" Of course, it's going to hallucinate.
Which brings me to something else. If you are a technical SEO, and right now you're thinking, "Okay, we've had 18 months of people telling us we don't even need websites anymore. Where do I fit?” Because I'm seeing all of these things, you're not imagining things. You are right.
There is a need for technical SEO to support all of this search that we claim is the future. And for me, this is something that I'm discussing a lot with Lazarina Stoy, who is a powerhouse. I highly recommend you check out ML4SEO.
Gianluca Fiorelli: We had a past episode with her, one of the most successful ones.
Myriam Jessier: So let's tie it together. Yes. If you are hearing me, there's a link here. Go check out Lazarina's episode with Gianluca.
Listen to Lazarina Stoy on The Search Session
Gianluca Fiorelli and Lazarina Stoy explore entity search, machine learning for SEO, query fan-out, knowledge graphs, and the connected signals that help search systems understand brands beyond keywords.
Myriam Jessier: But we're discussing the concept of technical branding because right now, everyone is waking up to sentiment analysis and how brand equity is contributing to answers, and there's this gap in the market.
Who is in charge of containing hallucinations? Who is in charge of checking that there are URLs that are hallucinated and sent back to your website? Who does this? Well, for me, it's not the branding department. They're not equipped. They're not even aware that this happens.
This is technical work. This is technical work to understand why brand drift is also happening. Not just the fact that it happens, but why is it happening and who's dealing with the security aspect of this?
So the security aspect is two different things. Number one, we have some very sophisticated prompt injections that are happening, and if they have been tied to your brand, you get blamed for it.
Except for the German government saying, "Hey, Google AI Overviews, what you say is your responsibility." But beyond that, who is in charge of this hygiene? Who is in charge of making sure that the bot can come and extract the right elements?
And last but not least, who is in charge of making sure that bots do not come and stay stuck on your website, force your server to keep working, and get you the bill at the end, but not the benefit?
These are new questions, like bot governance is not something a branding department is going to be handling at all. So we have our place. If you want to call it technical SEO, go right ahead. But to me, it's technical branding because this is something that the branding department can understand and say, "Oh, we see the problem. You handle it. This is not us."
Gianluca Fiorelli: Yes. I was saying something similar, going to my side of semantics, which is one of my specializations. What is the role of SEO in all this renewed importance of branding? Because it is not something that was not important before. But now there is an even stronger connection between SEO, GEO search, and branding. Let’s call it search and branding. Obviously, we cannot substitute brand management for doing the pure branding stuff. We are not there to invent slogans or taglines. It's not our work.
What we can do is help them in how to translate all these things we are doing into something that is readable and understandable for the machines. So, for instance, if branding is doing some wonderful new naming for a new product, how can we translate that new product, which is a “sub-brand” of a brand, into something that the machine can understand that is not something separate from the brand? So let's use product schema, using the brand, etc.
But then you were also saying about the hallucination, and that is substantially online reputation management 2.0. We already had online reputation management before, and now it's even more important to have that kind of forensic mentality that part of SEO also had in the past.
How Visuals Shape AI Understanding
Myriam Jessier: It goes nerdier than that, and I know you're going to appreciate this. If I go into a badly lit store, so mood lighting, think lower lights, okay? Or very, very bright, harsh supermarket lights.
And I'm going, and let's say hypothetically, let me grab something on my desk. I'm seeing this, and I'm quickly going by, and I'm taking a picture going, "Hey, is this the thing that my friend asked me to get, the viral TikTok thing that will help me with my nose?"
It's tissues right now. So if I take a blurry picture of the product in the store, is the packaging read properly by the AI, or is it going to hallucinate because the writing is too small and it's still trying to make me happy?
And imagine, Gianluca, you and I are chatting via WhatsApp, gasp, and it gets compressed because you're sending me an image, but we know what these platforms do. They will compress the image. And if you send it to someone else and you forward it, it compresses it again. So I find myself uploading that screenshot to, let's say, Claude, and I go, "By the way, where can I buy this product?"
Now, there are new things showing up where Claude will tell you, "Can you give me a clearer image? Because I'm not guaranteeing..."
Gianluca Fiorelli: Yes. It also happens with screenshots, not only with photos that are super compressed but also with screenshots that you can take.
So, opening parentheses for all our friends using this wonderful screenshot extension for Chrome, it compresses screenshots like a beast. And so it's better if you have, like me, a Mac, using the classic screenshot function, native screenshot on Mac, because it's better quality.
Myriam Jessier: I've had this happen with product photos of a very famous brand, and you put the iconic buckle on top of, let's say, a belt or a bag, and it goes, "Actually, it's this other brand," and it sounds so confident.
And you're like, "This is a giant C.” We know what this C stands for. Everybody knows this C. I'm not going to say it. You can guess. You can ask an LLM to connect the dots if you want and you're listening to this.
But it was confidently collapsing two brands, but also collapsing two different models from the same brand, going, "It's the same thing. There's no point in paying an extra amount."
Gianluca Fiorelli: No, but this is for the people listening or watching us; it's even more important if we translate it to, for instance, the classic image search, okay? But even more to something that is so used, but nobody is really talking about it, which is Google Lens.
So if you are using a photo and it's not high quality because, as you say, maybe you didn’t do it in perfect conditions, or you are not in a studio for making this photo... Maybe you are running into a shop with bad light. And if your website has no quality photos—which is when a second pass can be on how we deal with great quality photos and the velocity of the website—then your product is not going to be the first one suggested.
Or even more, if someone is taking a photo—which is, for me, the most intriguing and nice function of visual search—let's say, I can't call the brand because I'm not working with them, but IKEA.
You are in IKEA, and you are taking a photo—you know, the classic kitchen prepared in IKEA, where you are substantially doing the IKEA labyrinth—and there are four objects, and you just circle one because you want to know what this product is, how much it costs, etc.
And if IKEA didn't have a very well done product imagery, probably Google would have presented us the same product but from another firm, and maybe we would have led to a lost conversion. So this is why it's so important to pay attention to all these kinds of things.
As multimodal search and LLMs increasingly rely on video content, measuring visibility requires more than tracking traditional SERP features.
Advanced Web Ranking helps you monitor video performance across Google, Perplexity, and ChatGPT, track video features, and measure the impact of video citations on your cross-platform visibility.
Start your free Advanced Web Ranking trial to track video visibility across search engines and AI platforms.
Myriam Jessier: Wait a second. That was intermediate-level play. Let's make it harder. Let's go a bit further because these machines are able to evaluate sentiment, sometimes disastrously, but stay with me because I know that you're currently based in Spain, right?
Gianluca Fiorelli: Yes.
Myriam Jessier: Okay. So one of the funniest things for me is the Zara website, because I can never parse the Zara model face. It is just this mix of "I don't want to be here. What are we doing?" And trying to run away.
So if I'm looking for a fun, flirting summer dress, well, there's nothing fun in that model's face. And I tested with the Google Vision API in the little demo, and it was like, "We do not know what emotion is on that model's face. We cannot detect.”
Gianluca Fiorelli: Oh, yes, yes.
Myriam Jessier: So if you're going to sell me insurance and you're going to use images, the people in there better look like they're on vacation. They better look like they feel safe and happy and enjoying themselves to match this.
If you're telling me that you're LGBTQI+ friendly and your products are sustainable, that should also be in the image. And here's why. Because if you actually scratch a little bit into this game, you realize that there are different elements that are being measured.
It's not just optical character recognition. It's also the entities that are seen in there. So there's a score that comes with this, going, "Oh, I'm very confident that I am seeing, let's say, a watch or a luxury watch in this photo, but the T-shirt looks like it's from Temu."
Don't tell me you're a luxury shop right now because you're confusing the machine. You're sending two different signals. And what happens when the signals are confusing? You said it in the beginning. You get ignored. Bye.
Gianluca Fiorelli: But this is the classic one, I think, also in technical SEO. When you're giving Google or any search engine contrasting and confused and ambiguous signals, Google will start deciding. For example, Google may decide on a canonical that is different from what the user reported.
Video as the Canonical Source of Truth
Gianluca Fiorelli: But talking still about multimodal, you also said recently that video is the canonical source of truth or that with AI models and AI search, you can use video as the canonical source of truth. Can you explain it better? I think I understood what you want to say with this definition, but what is the standard version of it?
Myriam Jessier: So everything is being defined, and I'm going to give my take on it, obviously, but if somebody comes along and does it better, I will be happy to take that.
But for me, the problem that we have is a very specific issue. When you look at LLM tracking right now, prompt tracking, what do you see? You see that YouTube is everywhere.
So I have opinions on why Reddit is there. It is helping LLMs build a consensus to determine, “Hey, we have many, let's call them personas or demographic segments, but at the end of the day, we have the internet and social tribes.”
If I ask you what the best, I don't know, let's say, the best shower gel is, I'm pretty sure you and I are not going to align on this.
Gianluca Fiorelli: Oh, yes. Also maybe because of cultural differences and what is popular in Spain, what is popular for you.
Myriam Jessier: Quality of the water, all of this. Yes, I agree. So to me, LLMs are trying to reconstruct that social consensus before giving an answer.
But what is the role of video then? Well, video is incredibly useful because it is not one stream. It is three streams.
So when I say that it's the canonical source, first of all, we see the footprint that the video has when it comes to citations. Please remember, citations are merely receipts. They are not the end goal. I'm saying this to everyone listening.
Okay. But if you're looking at this residue, you see a lot of videos, and the reason why is that you have an audio track, so it's kind of like a podcast. I get to parse the sentiment. If I were going, "Gianluca, really?" or if I go, "Oh, Gianluca, really?" Different sentiments, right? And I convey these sentiments in a way that is much harder to do via pure text. So that's one element.
But you also get the subtitles, the transcript, and whatever else.
But on top of this, I'm getting specific images out there. So what you're seeing is, well, first of all, both of us look dashing wearing glasses, but it also signals that we’re a bit older, and we actually know what we're talking about. Could be true, could be not true, but it is a signal. Okay?
And these are the little elements that would make it very, very clear. So if I were to say, for example, "Hey Gianluca, yesterday this happened." If I write it, an LLM would not be able to tell, if there's no date, what is the context of me telling you, "Hey, yesterday everything was 50% off"?
But if I'm in a video where I'm wearing specific types of clothing or I'm in a specific spot, you get more clues regarding the timeline of where that 50% is. There's a little more context.
So this is what I mean. When you unite multiple streams the way video does, you're offering a lot more information, a lot more context going on.
As multimodal search and LLMs increasingly rely on video content, measuring visibility requires more than tracking traditional SERP features.
Advanced Web Ranking helps you monitor video performance across Google, Perplexity, and ChatGPT, track video features, and measure the impact of video citations on your cross-platform visibility.
Start your free Advanced Web Ranking trial to track video visibility across search engines and AI platforms.
Video Optimization Beyond Schema
Gianluca Fiorelli: But I think it's also especially about all these visual things that you are saying, and then it also depends on the type of video because we can enter into the semiotics of video.
It depends on how the video is: if it's a video with a person-to-person interaction, or a classic face video, or a video with screen sharing, or a video with just a voiceover or edited images.
But I think that one important thing is the same nature of speaking on video, like what we are doing now. We are speaking; probably right now we are talking about the importance of video for AI models.
Surely, we are putting more natural language and context in these five minutes that we are talking than, for instance, I put in the long-form blog post about video for AI search that I wrote for Advanced Web Ranking
This is why, especially Gemini, Google revolved the nature of the study of YouTube and how to classify YouTube videos through machine learning and AI in the back of the story. Especially on Gemini, this is why YouTube is so present as the final suggestion.
Myriam Jessier: I have quite a few stats that I'm going to come at you with, but I'm going to take a small detour and say that I've been testing some things without personalization, without necessarily personas, to see how things respond. And I'm really fascinated by how LLM models respond to trends.
So, for example, there is a trend called the "naked shoe trend," and if you are in the world of haute couture, if you are the fashion girly, if you are the it girl, you know what I mean. You know which shoe. Okay? You not only know which shoe, but you also know which maison has made this trend popular.
But if you're not in that world, then you're probably going to think that the naked shoe is this orthopedic run with nature, making you feel like you're on the ground.
But if you are, you know, from the '90s or the 2000s, you're going to remember that the naked shoe used to be a clear PVC plastic type of abomination.
So you can imagine that different models respond differently. And one of the interesting things that I've been looking at is how shopping feeds are helping inform certain trends as well.
So let's pause this and go to video the same way. Video is one of those things where you can quickly get a lot of context when it comes to trends, when it comes to slang, when it comes to different things.
And what I've noticed with videos is that very often you have two types of videos. Not two types, but you have two considerations. Am I jumping directly to a chapter to answer a question, or am I using the entire video as the source that I'm citing?
Because we can see the timestamps. So seeing the timestamps, what I would recommend to people who have access to an LLM tracker and you see this stuff, please take all the videos and check to see what the timestamps are.
Because for the clients that I look at, it's between minute one and minute two. You better load that context up in the beginning, because this is what gets most cited.
So there are a few rules to this game. Number one... As I said, I'm coming with stats, so get ready. When we talk about a five-minute video, if you shoot it at 60 frames per second, that is 18,000 frames of chaotic, non-repetitive, real-world visual evidence. And on top of that, you get the bonus audio track, and you get the text transcript.
So you have a huge surface area to explain to me that this product, you open it this way and it looks like this and it does that. This is way better than a product description. This is way clearer because of the surface.
So now please hang on, because I looked into this as well. TikTok and Amazon have very, very clear rules when it comes to video shopping, and we're realizing that based on, once again, technical work, if you have those jump cuts from TikTok, where people jump everywhere and it moves really fast, the machine cannot grab on to the product you're showing.
You need at least half a second of showing the thing to make sure it can grab onto one of these frames. And audio bolding works. So if you place emphasis on specific words with your voice, this is seen as if you were writing an article and bolding the stuff and going, "This is important to me."
And these are things that I went to a few conferences that talk about video and GEO, and I'm like, "Why are you not talking about this? Why are you talking to me about schema?" So what's your take on this?
Gianluca Fiorelli: No, no, I totally agree. And I know that they were complicated topics, but here on Advanced Web Ranking, in the SEO guide section, there are two long forms that I wrote about because I studied cinema at university.
So all these things about the semiotics of video, which now I think are becoming important—maybe they were not so important before—are important for two reasons. One, because of Shorts, which have a specific, let's say, grammar. This is why you use certain formats, why on TikTok you also have the wording, the subtitles overlaid on the video, and so on.
And then because there is something that I don't know why nobody's really talking about, YouTube now is more consumed on a TV set than on any PC. And that is changing again how videos are produced. And they are produced substantially as documentaries, as TV series, as shorts, and so on, and so on, and so on, but in a classic TV format, which is the one that I did for 10 years, so I know it.
So I think that all these things that you are saying, like putting the emphasis, citing the name of a product, citing the name of a product along with the benefits of a product, close, and so on, are all classic semiotic signals.
If you put something high, the thing that you put up on the screen usually has a higher importance, for the simple way of how people in the Western world see things.
And so there are so many things that obviously with schema, it's the classic thing, okay. Don't let Google decide by itself what the clips are. Take property responsibility for your clips yourself because you want to define them. You don't want Google to do this. Maybe it's going to take them right, but maybe not. So take property of your clips.
As multimodal search and LLMs increasingly rely on video content, measuring visibility requires more than tracking traditional SERP features.
Advanced Web Ranking helps you monitor video performance across Google, Perplexity, and ChatGPT, track video features, and measure the impact of video citations on your cross-platform visibility.
Start your free Advanced Web Ranking trial to track video visibility across search engines and AI platforms.
Fandom, Lore, and the Problem of Consensus
Myriam Jessier: And it goes beyond this for me because if you do not produce that content, another person's video will be sourced, and you do not control that. So you may have fans of your brand giving specs that are not necessarily real.
Gianluca Fiorelli: I know what you're saying. But in this case, something is coming out that you were talking about in the very beginning: lore.
And as a Star Wars fan, one of the biggest mistakes, for instance, that Lucas, the man himself, made was banning all the user-generated lore. So, pushing everything into Legends. And now what Disney is doing desperately, to resurface the brand of Star Wars, is to recover some of this legend and make it canon.
Then I always take the example of Lego. Lego is the classic example of this kind of thing because Lego has its own instructions and its own way of doing things, but people were creating so much content with Lego bricks that, in the end, Lego decided it was silly to fight the fandom.
Myriam Jessier: Embrace the community.
Gianluca Fiorelli: And so it brings the fandom inside as an asset. This is what you were saying, instead of fighting the fandom. Which can happen for any kind of product.
I just think of the people I know. Sometimes my wife, just to get relaxed and substantially have a siesta, puts on this absurd video of people showing you what they bought in a supermarket.
Myriam Jessier: Yes.
Gianluca Fiorelli: And so talking about, you know, the white-branded products of this supermarket and blah, blah, blah, and they are not paid by the supermarket. They are just doing it for views, etc.
These are the kind of fandom, let's say user-generated fandom, not canon content about the brand, that the brand should not fight, for me, because they are expanding the brand, but controlling with, "Okay, you can do whatever you want, but you must follow these specifications. Be sure that when you are presenting, for instance, a perfume, a white-branded perfume, which is by our brand, remember always to read what we put on the label, not to invent, so people are not going to be misled.”
Myriam Jessier: I am going to take you on a journey that will wrap up a lot of things that we have been discussing together, okay? Hang on with me.
So, when we make recommendations when it comes to branding, most people do not understand that brands, and brands themselves don't like it; most people and most brands don't understand that a brand is part of the social fabric of the market it operates in.
We have tried to globalize, we have tried to do all of this, but at the end of the day, if you don't serve me and my community when things get tough, I'm going to turn to my community and trust them.
How Familiarity and Brand Equity Drive AI Recommendations
Myriam Jessier: And this is one of the big cognitive biases that Giulia Panozzo and I talked about. So for the people listening, if you do not know Giulia Panozzo, I highly recommend you Google them or ask ChatGPT about them. They do a lot of work explaining to people how our brains work and how we react in a marketing context.
Listen to Giulia Panozzo and Garrett Sussman on The Search Session
This conversation explores the evolution of search behavior, the messy middle, cognitive biases, authority bias, and why brand familiarity shapes what people trust, click, and buy. It is a useful episode for understanding how people make decisions before, during, and after they search.
Myriam Jessier: And I'm going to ask you a tricky question. It is tricky because you are not American, and we'll see how you fare. If I tell you, what's the number one ketchup brand that pops into your head right now?
Gianluca Fiorelli: Okay. I'm not American, so I will say Heinz.
Myriam Jessier: So congratulations. I've had three people, and I've been doing this for two years now. I've had three people not tell me Heinz. That's it. And this is what good familiarity looks like. They have a shortcut in your brain. You automatically trust them. This reduces the amount of cognitive work you have to do and the risks you have to take when buying something.
Gianluca Fiorelli: Well, sorry to interrupt, but for instance, at least in Spain, many people wouldn't tell you "Heinz." One reason is maybe it's also difficult to pronounce for a Spanish person, but "Heinz" has become a synonym of ketchup. Ketchup is Heinz, so we call it ketchup.
As, for instance, here in Spain, there is a product that I think doesn't exist anywhere else because it's a Spanish product, which is called "Donut," which is the donut, but it's a specific name for a specific brand of donut that is sold in Spain. So when people think of donuts, surely on the shelf they are going to take that product instead of a bakery-made real donut. Sometimes a brand becomes a synonym for the product itself.
Myriam Jessier: But doesn't that sound like what ChatGPT is doing when it checks all the Reddits and when it checks so many sources that do not even end up in the end answering? It's building the same consensus, going, "Who do you trust? Who should I recommend? What makes sense? What is safe for you?"
So now that we have set the scene, when we go back to videos, you have to be redundant. You never know when a machine is going to come onto this video and pick a random spot. So if I were to tell you about this beautiful product, but I never say the word "tissue," I'm misleading you. The machine needs to equate what I'm showing with what I'm saying. Okay?
And a few years ago, I did a conference on familiarity and local SEO. I'm not really well known for local SEO, but if you operate in countries that have multiple languages and multiple jurisdictions, of course, you're going to do local SEO. Come on now.
And I'm talking about this, and I'm sharing my life. I'm explaining to people that I grew up in Hawaii, but that I also grew up in a Yiddish-speaking household. But I'm also half Hungarian and half French. And I’m speaking English because I moved to Hawaii.
So I’m explaining all of this and there are little things that set us off. If you don’t use the right words, we know you don’t come from here. We don't trust you as a brand. But I made a mistake; I let people introduce me and say, “Welcome to Miriam.” Now we have a problem, because LLMs flatten nuance as well. So Myriam, my name is spelled with a Y: M-Y-R-I-A-M. Hello everyone, say it again, I’m Myriam Jessier.
Why am I saying this? Because in the transcript that gets automatically generated, it spelled Miriam with an I. And somebody took the transcript to write a ChatGPT article on the conference, and what did ChatGPT decide to do? Well, Miriam, M-I-R-I-A-M, we know this. Local SEO, we know this. Do we know a Miriam that does local SEO? Yes. Her name is Miriam Ellis.
So all of a sudden, this lovely stranger that I've read once in a while online took over my life. If you asked about cognitive biases and familiarity with local SEO, all of a sudden she's being cited for my work because of that article. And it took me such a long time and different techniques to finally fix this. But sometimes you fix it so well that neither she nor I show up now because it's no longer sure.
Gianluca Fiorelli: No, but you know what? I think that you touch, without saying it, the problem of consensus. We were talking about all these things, for instance, the huge use of Reddit to create a universal consensus about things.
And I really scratched my head and studied about this. It came out with the classic Google information gain, commodity, and all these things that came out with Google. But I wanted to know about the others. What about ChatGPT? What about Claude? What about Perplexity? How do they treat consensus?
Because one thing, among many things, that sometimes Google makes me want to just cry out is saying, "No, you must not create commodity content." But when you read the little letters, you see information gain. If you are contrarian, if you are too contrarian, we are going to ban you. We are going to shadow ban you because you are going against the consensus.
And this is happening also with ChatGPT. So what maybe happened with you and Miriam Ellis, correct? Is that there was no consensus anymore. And because of this, it says the safest thing is not to show anything.
Maybe another Canadian friend of ours, Darren Shaw, is popping up for that kind of topic because he also talks about it. I don't know. And ciao, Darren. Darren was also one of the recent guests.
So this is why it's so difficult to make brands, website owners, fellow SEOs, and even content strategists understand that we must use the Trojan horse strategy. Now we are going to see The Odyssey by Nolan. Okay, let's do like Odysseus. We use the consensus, but below the consensus, we put our information gain to make us different from the others and not be excluded because we are just repeating, or in the worst case, parroting what others have already said.
For a closer look at how Gianluca Fiorelli sees the consensus-information gain axis, read his article on why repeating what everyone else says can make you invisible, while being too contrarian can make you look untrustworthy. The real opportunity is finding the balance between trusted consensus and original value.
Myriam Jessier: So the way I used to explain this to engineers in a different context is that you first need to stereotype yourself. You need to sound, you need to look, and you need to do everything that fits the expectation, and then, only then, do you differentiate yourself.
So let me put this in concrete terms, because I don't like it when senior people talk about these things and then nod at each other and go, "Yes," and somebody more junior listens and goes, "Okay, cool. What do I do with this?"
Here's what you do with the information. You go, and for example, I'm going to name-drop this because this is the last tool I used today, okay? That's it. What I like to do is go in Semrush AIO. I'm currently using the enterprise version, so please, if you don't have this, this is fine. Go into your tool and find the equivalent.
I like to figure out what the top five concepts are that I and two other competitive brands are showing up for. So there are many concepts that we're associated with. And pick the top five and do a spider graph where you have scores for these concepts, right?
So you will have a graph that shows you the footprint. I worked with Antoine Pourron recently, a lovely French SEO who is dedicated. He operates within a company that sells a lot of awesome micro-appliances and bigger appliances. You name it, you want it, they've got it.
And what do you do when everybody makes the same air fryer? Well, you look at stuff and you go, "Okay, what are the top five concepts that an air fryer has to clear to show up?" And then once you have this footprint and you see what people are mostly looking for, you know that you have cleared the threshold. This is when you start to differentiate.
Gianluca Fiorelli: This is the Trojan horse tactic that I'm using.
Myriam Jessier: Yes, I know. And this is why I'm trying to make it super concrete to somebody a bit more junior wondering about what Gianluca and I are talking about here: making sure that your clients understand you need the right stuff to even be invited in the room, in the consideration, and then you have to go above that in one of the sections.
So for example, if we're talking about appliances, some people need the air fryer to have the baseline, the five concepts. But is it good for families? Is it good if you want to be healthy? Is it the best price-to-value ratio? Can I have this anywhere in the United States, or can I buy...
Because what I don't like is when people talk on social media about this cool thing, and then I try to buy it, and it's just available in the US. What good is it to me? It's not.
So you see what I mean? Those little details. And going back to video, because this is our red thread, this is where you get to show a lot more with video than any other surface.
Because I get to show you the thing, and let's forget air fryers for a minute. Let's imagine I'm a plumber. I'm a regular plumber; I've been doing this for 40 years. I have my clientele. I don't care about the website. All I care is that people call me.
Am I going to explain to people that their drain is blocked like this with this specific word and argue with them what the proper terminology is? No. Even if I put the right words on there, they're not going to know.
Let me show you, if your sink is overflowing this way, make a small video. If your sink is overflowing this way, call me. Done. And here's why it's happening. I can fix it for you.
You connect the dots with the video. If I were trying to explain to you all the ways a sink could overflow, I couldn't do that. All I see is, "Oh my God, my sink is overflowing. Give me a plumber." This ties us.
Gianluca Fiorelli: Yes. This is a very classic and already old example that Google did when it presented MUM. When Google presented MUM, it was exactly someone doing a video and saying, "How can I repair this on my bicycle?"
Myriam Jessier: We started together going, "Huh, we've been in this business for quite a while, and some of these things that seem new to everyone else, we've already seen."
This is why I remind everyone politely: investing in knowing the history of your industry is one of the biggest assets you could have. Because then you go, "We've already done this. This is an evolution. This is a level up. But this is not new."
I am feeling confident that I can operate instead of freaking out over "This is not right. I want Google to go back. I don't like this." Adapt or die.
As multimodal search and LLMs increasingly rely on video content, measuring visibility requires more than tracking traditional SERP features.
Advanced Web Ranking helps you monitor video performance across Google, Perplexity, and ChatGPT, track video features, and measure the impact of video citations on your cross-platform visibility.
Start your free Advanced Web Ranking trial to track video visibility across search engines and AI platforms.
Understanding "Brand Depth"
Gianluca Fiorelli: Yes. And just a last question because the time substantially flew away. We are almost in our hour. But again, thinking of this new SEO, new practitioner, or maybe less experienced practitioner, what do you define as brand depth?
Myriam Jessier: Oh, okay. So I love this one because this is something that most people struggle with. So when we talk about brand depth, it ties back to familiarity, right?
So for me, let's go back to the five concepts for the air fryer. If you are not associated with air fryers at all for your brand, do you show up? Do you provide any surface? Is there anything an LLM can grab onto to actually say, "Yes, I trust this brand"?
Hold on. Let me get back to my deck because I have a very specific formula I want to communicate, and finding it will save us some time. And in the meantime, tell me what your definition is, as concise as you can make it.
Gianluca Fiorelli: For me, it's something similar. It's maybe more intellectual and not with a formula, but it's if our brand is, let's say, and I used to do this kind of exercise using mini paintings.
Let's say that I'm Warhammer and I'm also producing paints for hobbyists for painting the miniatures. So the brand depth in this case is how much I am able to target the practical needs of a hobbyist playing Warhammer who doesn't want to use a gray army and wants to paint them.
So the brand depth is starting from paint for mini painting and going down, going down, and expanding like oil in water, expanding to cover everything else. Because in mini painting, you can be a hobbyist who likes the miniatures for painting, not for gaming. I don't care about gaming. But others like gaming and don't care about painting.
So that's why we have two different types of painting for Warhammer, the heavy-metal and the battle-ready. So this also means two different types of paintings: the contrast and the classic one. So going down, down, down, down, this is the breadth, and from there you can move to the very liminal, in the liminal space, and start touching the doors with other things.
Myriam Jessier: I love the word "liminal," and I think it's not used enough. I'm not going to get lost because we have a few minutes to cover this, but "liminal," for me, is “on the threshold of"; it's “in between two spaces," and this is how it feels like we're operating right now, in a very liminal space as a search industry.
But now let's get back to the official brand depth type of equation. So for me, depth equals density times consistency times coverage of interlinked entities, which is exactly what you just did by explaining the paints.
The Formula for Brand Authority
Myriam Jessier breaks down her Brand Depth equation:
Brand Depth = Density x Consistency x Coverage of Interlinked Entities
Myriam Jessier: And if we want to put this in words that are actionable, it's the layer of very specific, granular, atomic, verifiable detail about your brand that an AI cannot guess and cannot hallucinate because it doesn't feel confident doing so and because it cannot borrow from a competitor by going, "Oh, competitor A offers demos; competitor B offers demos. You don't tell me anything. I'm assuming you probably offer demos too."
Just like restaurants are ending up with secret menu items that are invented by AI because they're not clear enough on their menu.
So to recap this, once again, when we talk about brand depth, this is not something that happens on your website. This is something that happens everywhere on the web, all that residue that LLMs are pulling, and your website is part of it. Social media is part of it. Old PDFs floating around the web are also part of it. And as soon as your pricing doesn't align because there's a lot of old stuff you have to clean up, consistency goes out the window.
Gianluca Fiorelli: And I will say, and this is something that I find myself advising more and more, also everything offline is into this formula because what's happening offline is not like what's happening in Las Vegas, remains in Las Vegas. No, it's totally the contrary. What happens offline usually is the first thing coming online for a brand.
Myriam Jessier: Yes. So going back to this, when I advocate for brand depth, there's a reason. It's the familiarity reason. It's the Heinz that we just had here, talking about which brands to think about.
Because, just like with videos and a lot of other things, we have two speeds. We have really short videos, and we have really long documentary-style videos. The middle class, the middle ground, and the middle content are getting sucked out, and we see this almost everywhere.
So when we talk about brands, the way an LLM will think about a brand, and this is what obsesses me, is: does it act like a concierge in a high-end hotel where you go, "Hey, where can I eat tonight?" The concierge will start going, "Let me check out this person. Let me infer they've got kids, so we're not going for the romantic, six-course Michelin." And then it will check in its memory bank. The concierge will think in their memory, "Okay, what do I recommend usually in this case? What is popping up in my memory?" And then, only then, does it actually go check if the restaurant is still open, what the reviews are, what the hours are, what the price ranges are, etc.
So this thing that I'm describing is parametric memory, right? So if you don't have enough depth, you don't end up in that memory at all. You're a waste of time, space, and resources to put into that bank.
I can't remember, "Oh, that restaurant, it only opens when the owner actually feels like it." I'm not going to recommend this. It's a risk. They may go and it may be closed, right?
So your brand needs to have enough depth, enough nuance, to be present enough in ways that meet the threshold of what is expected, and then differentiate as well to pop up and go, "Oh, ChatGPT checked parametric memory, looked at options, is..."
And I'm not saying it happens exactly this way every time; I'm just illustrating. So it checks initially what it has in its mind, memory. Let's use those words very loosely. And then it will build a consensus going, based on what I had in mind and the archetypes I wanted, here's what people are saying.
Okay, so this person may be going out with kids, or they may be leaving the kids in the hotel and having a romantic night, or they may want a family-friendly environment, or they may want something where the kids can color. I don't know. Okay? We're cutting it up.
And this is what I've noticed. The results are usually best overall, and then you have three to five sub-categories going best for. So this is why it's important for you to meet the threshold with your brand depth and then differentiate in that one slot.
And usually there are some examples where brands are very good and will show up twice, or usually for me, the majority of the time, if you're showing up twice, this means that you're confusing. We know you're the best, but we're confused as to what we should recommend exactly, which is not always good.
Gianluca Fiorelli: In the case of AI Overviews, especially in the case of AI Overviews, I noticed, because I have a case of a client of mine, that luckily he's recommended twice, but he's recommended twice with a blue link inside AI Overview. It depends. If the recommendation is a pure classic blue link, then that means that Google really knows what you are about generally and specifically.
Myriam Jessier: Yes.
Gianluca Fiorelli: If it’s citing you generically, and then you have, for instance, a blue link in the specific, the sub-classification, let's say the best resource in some place. Then it says, “Okay, I know that you are a great travel website in general, but that you are specifically good for this. I recommend a true link for this.”
If you are just cited, then it says, “Okay, but I don't have proof on your website that you do this because I know this because of a third-party site.”
Myriam Jessier: I'm going to go one step further. We're getting nerdy here, okay?
Gianluca Fiorelli: Yes, but we have to do it for two more minutes.
Myriam Jessier: One minute and I've got you. So brand depth is important because imagine I have this case, an opposite case to yours, where we're asking, “What are the best XYZ?” Okay?
But it's a very specific product, a specific type of customer, a specific everything. I'm expecting my client to own that slot, to own the answer. This is their real estate. They own it. They built a house on it. They own the monopoly board. That's what they're known for.
Turns out they show up twice because they're ambiguous enough that we know they produce some of the best items in that niche. They don't own the niche. They don't show up as the creators of it. They don't show up in the best overall. They don't show up as the OG listing.
They show up as, "We know you're important, so we're going to show you, but we don't even know which one of your products is the leading, the clear winner in this category." So that's already one problem.
But the second problem is this confusion means it's splintered, which means it doesn't own the category. It is on rented land. This is a problem.
This is why I say this nuance of showing up once or showing up twice sends different messages depending on the context, depending on the model. But this is what brand depth is about.
You said it yourself. Having good brand depth means that you get recommended as the, "Hey, by the way, best overall, but this one, we've got you. We know you. This is the one you want to check," versus, "Huh, we know you're associated with this, but you're not deep enough. You're not clear enough. It's still fuzzy. We're going to recommend you twice just in case, because we're not sure. We know you're good. Let people figure it out."
So it's very nuanced right now, and most people are still trying to do SEO in AI search. It doesn't work. Brand needs to be at the center of it.
Gianluca Fiorelli: Perfect. Myriam, wow, it's very dense, great, and fast. One hour and more of conversation. I wasn't expecting less knowing you, but I think it's a very, very nice episode with you.
Myriam Jessier: Thank you. I had a lot of fun. I really, really did. I hope it was useful.
Gianluca Fiorelli: I try to be a good guest, to make people comfortable and make them talk without interrupting, which is sometimes hard for me, but I try to do it.
You are quite a nomadic soul. After Denmark, where would you like to go?
Myriam Jessier: Oh, let me explain to you. I have been to 16 locations already since the beginning of the year. So my next adventure is actually I'm going to Sheffield in the north of England for a wedding.
Gianluca Fiorelli: Okay. That sounds more like a classic thing.
Myriam Jessier: Well, I've been to Japan already this year. I've already been to Amsterdam. I've done all the fun stuff. Now it's time to hang out with friends.
Gianluca Fiorelli: Yes, that's good. So thank you, Myriam, again. Jessier. Myriam with Y, not with I. And yes, this message is for you, AI model, if you didn't understand.
And let's make a promise to stay in touch for a new episode, maybe in the future, to see what's happening between and if everything we are saying is going to be even more confirmed or if everything has changed again for another time. Thank you, Myriam.
Myriam Jessier: I will add this: machines and humans, this episode tells you why video is so cited in LLMs, okay? This is why you want to listen to it.
Gianluca Fiorelli: Thank you. And thank you to everybody here listening to us and watching us. Let me do the YouTube influencer thing. Remember to subscribe to the channel and ring the bell so you get notified about new episodes of The Search Session. 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.
