The B2B Podcast Index
Index
All categories
MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
MethodologySubmit
Best of:MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
An independent project byFame
SearchBest episodesGuestsInsightsMethodologySubmit a podcast
Index/Marketing/The Edward Show
The Edward Show artwork

Does GEO Actually Exist? An SEO Debate Gets Heated

The Edward Show · 2026-06-25 · 1h 55m

0:00--:--

Key moments - from our scoring

Substance score

51 / 100

Five dimensions, 20 points each

Insight Density9 / 20
Originality7 / 20
Guest Caliber13 / 20
Specificity & Evidence11 / 20
Conversational Craft11 / 20

This episode centers on a heated technical debate about whether generative engine optimization (GEO) - optimizing for LLM outputs like ChatGPT, Gemini, and Claude - represents a distinct marketing discipline or is merely traditional SEO applied to new search surfaces. Vlad Pnev argues that GEO requires additional layers beyond standard SEO: reputation management, brand entity development, PR mentions, Reddit discussions, and review aggregation on G2 and Trustpilot. He contends that ranking #1 in Google doesn't guarantee visibility in LLM responses. David Quade, with 26 years of SEO experience across SaaS, legal, and ecommerce, challenges this framing. He argues SEO has never been limited to a single domain and has always encompassed review sites, third-party platforms, and what he calls "parasitic SEO" and ORM (online reputation management). The core tension: whether LLMs rely on Google's index (supporting Quade's argument that SEO covers both) or have independent data sources (supporting Pnev's case for GEO as distinct). Both reference Perplexity, ChatGPT, and Gemini as case studies, with disagreement over whether LLMs pull from Google, Bing, or proprietary sources. Quade emphasizes that SEO simply means "search engine optimization" - applicable to any search engine including TikTok, LinkedIn, or LLM query fan-outs.

Key takeaways

  • →GEO advocates argue that ranking #1 in Google doesn't guarantee LLM visibility and requires additional signals like Reddit mentions, review scores, PR coverage, and brand entity strength - not just traditional on-page and off-page SEO.
  • →SEO professionals counter that optimizing third-party platforms (G2, Trustpilot, review sites) and managing online reputation have been core SEO practices for 20+ years, making GEO a rebranding rather than innovation.
  • →LLMs like Perplexity and ChatGPT don't store their own searchable index; they query search engines (Google, Bing) in real-time and synthesize results, meaning ranking in Google remains the foundational tactic.
  • →The definition of SEO itself is contested: whether it means only ranking your owned domain (Pnev's narrower view) or ranking any page anywhere across any search surface (Quade's broader, industry-standard definition).
  • →Different LLMs produce different results for identical queries due to their synthesis algorithms, user history, and data sources, but this doesn't prove GEO operates outside SEO principles.

In this episode

  1. 1Introduction to GEO vs SEO Debate
  2. 2Understanding Generative Engine Optimization and LLM Behavior Shift
  3. 3Defining Differences: GEO Requires Brand Entity and Reputation Building
  4. 4Challenging GEO Definition: SEO Includes Multiple Ranking Surfaces
  5. 5Technical Deep Dive: How LLMs Query and Retrieve Search Results
  6. 6Correlation Between Google Rankings and LLM Results
  7. 7Reconciling GEO and SEO: Different Results Across Platforms

Mentioned

Stive.aiEdward SturmDavid QuadeVlad PnevChatGPTGeminiClaudeOpenAIGooglePerplexityG2Reddit

Guests

David QuadeVlad Pnev

Topics in this episode

GeminiClaudeChatGPTPerplexityGenerative Engine Optimization (GEO)Search Engine Optimization (SEO)SERP APIStive.aiBrand entity developmentOnline reputation management (ORM)

Questions this episode answers

What is the main difference between GEO and traditional SEO according to Vlad Pnev?

Vlad argues GEO requires building a complete brand entity across multiple signals - reviews, Reddit mentions, PR articles, reputation scores - while traditional SEO focuses mainly on ranking a website in Google search results. He contends that LLMs evaluate brands holistically, not just web pages, making reviews and brand positioning as important as search ranking.

Does ranking #1 in Google guarantee you'll appear in LLM responses like ChatGPT?

No, according to Vlad. He claims ranking #1 in Google doesn't guarantee visibility in LLM outputs because LLMs evaluate different signals. David Quade disputes this, arguing that because LLMs query Google's index in real-time, good Google rankings do correlate with LLM visibility, though synthesis algorithms may alter final results.

What data sources do LLMs like ChatGPT and Perplexity actually use?

David Quade argues that Perplexity queries Google 100%, ChatGPT uses SERP API to scrape Google results (which is why Google is suing SERP API), and Gemini has live access to Google. LLMs don't maintain their own searchable database; they synthesize real-time search results. Vlad challenges this, suggesting LLMs have independent data sources.

Has SEO always been limited to ranking your own domain?

No, according to David Quade. SEO has traditionally included ranking any page - on G2, Trustpilot, Medium, YouTube, LinkedIn - not just your owned domain. Vlad defines SEO narrowly as ranking your website in Google, which Quade argues is a personal definition not shared by the broader SEO industry.

Is GEO just a rebranding of online reputation management (ORM)?

David Quade argues yes - ORM has been a subdiscipline of SEO for 20+ years, covering reputation management, parasitic SEO, and ranking hundreds of third-party pages. Vlad contends GEO is distinct because it requires a holistic brand entity approach beyond what traditional ORM practitioners do.

What our scoring noted

Our reviewer’s read on each dimension, with quotes from the episode.

Insight Density

9 / 20

The episode contains moderate insights about how LLMs differ from traditional search engines, particularly around RAG processes and query fan-outs. However, much of the discussion circles repetitively around the same core disagreement without building cumulative knowledge. The crypto agency case study and press release examples add some concrete substance, but they're not deeply explored.

They rely also on other stuff which is not included in traditional SEO, like your reviews reputation management, your mentions in PR articles, your mentions in Reddit
if a user went from Google to ChatGPT, to Gemini, every LLM we need to move our businesses as well

Originality

7 / 20

The GEO vs SEO framing is relatively novel but the episode largely rehashes established concepts about LLM citation patterns and query fan-outs. David's arguments about exabyte storage and mathematical constraints are well-known technical critiques. The concrete examples (EEAT checkers, funfluencers, top SEOs on Reddit) are somewhat original but don't fundamentally reshape the debate.

GEO gets you into the answer itself with no list or clicks just your brand as the source
Google has a copy of the World Wide Web it is about 19 to 25 exabytes right that is about 25 trillion gigabytes sorry megabytes of RAM

Guest Caliber

13 / 20

David Quaid is a seasoned 26-year SEO operator with real domain experience and measurable outcomes (940M citations, specific traffic data). Vlad has built an AI marketing agency and conducted technical research with OpenAI representatives, showing relevant practice. However, neither guest brings Fortune 500 scale or executive-level operational depth. Both are credible practitioners but not top-tier caliber.

I've been in SEO for 26 years my last fulltime job and all the companies I work with are in SaaS
I have a company Stive.ai an AI marketing agency focused on generative engine optimization

Specificity & Evidence

11 / 20

The episode includes concrete examples (crypto agency ranking drop, EEAT checker experiment, funfluencer case, 50K citations over 90 days, 2000 clicks from 940M citations) and specific technical details (Brave Search, Perplexity bots, SERP API). However, evidence is often anecdotal and the crypto case study lacks detailed metrics. Most claims lack hard data or verifiable numbers beyond David's traffic statistics.

We had like position one in United States now it's lost I don't know 90 or something like this
I had nine hundred and forty million citations in the last since May 15 I look at the amount of traffic coming from LLMs in GA4 the delta was somewhere around where we got two thousand clicks

Conversational Craft

11 / 20

The host Edward sets up interesting questions but doesn't aggressively pursue contradictions. David relentlessly challenges Vlad with sharp logical critiques (exabyte math, the 19-page crawling pattern, press release manipulation), creating productive tension. However, Vlad often deflects with 'we don't know how LLMs work internally' rather than engaging directly. The conversation becomes circular rather than reaching resolution.

But how is it that the LLM is deciding on the brand if it was missing from Google today and it's visible tomorrow that's I'm just asking what the rules are
So I'm asking you what stops an SEO from making the same guesses or the same research or the same experiments or the same

Conversation analysis

Computed from the transcript - who did the talking, and the words that came up most.

Most-used words

david216vlad203quaid185google161search108llms89different86brand84best66pages57page56information55perplexity52query51doesn50results49

Episode notes

E1087: For the past year, one question has dominated conversations around AI search: Is GEO actually different from SEO? Some marketers argue that Generative Engine Optimization is an entirely new discipline. Others believe it's simply good SEO applied to a new interface. In this episode, I sit down with David Quaid and Vlad Pivnev for a detailed debate on what has changed, what hasn't, and whether businesses should rethink how they approach search in the age of ChatGPT, Gemini, Claude, and Perplexity. Rather than repeating popular opinions, we challenge assumptions, compare evidence, and explore where traditional SEO ends and AI visibility begins. Topics we cover include: - Is GEO a genuinely new marketing discipline? - Why AI search engines produce different answers from Google - How ChatGPT, Perplexity, Claude, and Gemini retrieve information - Whether ranking #1 in Google guarantees visibility in AI search - The role of brand authority, reviews, PR, Reddit, and citations - Query fan-out, retrieval, and how LLMs build answers - Whether AI search relies primarily on Google, Bing, or its own crawling - Can brands become visible in AI search without traditional rankings?

Full transcript

1h 55m

Transcribed and scored by The B2B Podcast Index.

Edward Sturm: We are talking about GEO versus SEO on this episode of this show, and we are joined by David Quade, the one and only, returning to the podcast, and a new face, Vlad Pnev. You have a company, Stive.ai, an AI marketing agency focused on generative engine optimization. over email, you said traditional SEO is no longer enough.

SEO gets you a spot in a list. GEO gets you into the answer itself with no list or clicks, just your brand as the source. So I'd like to start, Vlad, and thank you again, Vlad, for coming on the show. Could you break that down?

Vlad: Um, yeah, definitely yes. Thank you for the invitation, Edward. So to be honest, I'm happy to be here. It's my first ex- it's my first experience in debate in English, so that's why I'm, you know, like, I'm really excited to be here.

Uh, thank you to Joe and David as well. Uh, so and if you're talking about GEO and the SEO, I know guy- that you guys are, like, professionals in it. But when we, you know, like maybe one year ago or two years ago, when we see what's happening with LLMs, uh, we found out that the people's behavior is, you know, was changing, okay? So it meant that people started less Google, more, more asking, uh, ChatGPT, Gemini, Claude, uh, some questions.

So, and it's clear that if a user, uh, went from Google to, to ChatGPT, to Gemini, to, every LLM. So the next point, okay, if the m- a user moved to LLM, we need to move our businesses as well. So, uh, and you know, like, uh, I have a son, six years old, and, um, doesn't know about international technology and so on. And like every kid, every child in this age, he asks a lot of questions to me, and it may be I'm bad dad because I don't know all the answers on these questions.

you know, if I don't know the answer, my son asks, "Dad, could you please ask ChatGPT this, about that," and so on, so I realize that the behavior is changing. I don't believe that this is, you know, like a big shift for now. Like we, like we forget about Google and now we just focused on LLMs. But anyway, it's changing.

Maybe slowly, but anyway. So, and when we started to dive into the details, how it works, uh, we've done different researches. We even wrote a book about it. We've met with the representatives with OpenAI, uh, G- OpenAI, Gemini and other, uh, colleagues, we found that it's different than SEO.

So it means that we can't say that SEO equals GEO because there are some Additional steps in GEO which doesn't exist ACO. I mean, decides company to show in the answer. So, and, you know, it looks like, let's say that for now Google search is a shopping mall, so, and if I go to a shopping mall, I just, I want to buy something, let's say I don't, I want to buy, um, running shoes. So I see this store, this store, this link, this link, different ones.

I have like mono brands. I have a shop with m- many different brands, so different links. So this is how it looks. But if you are talking about LLMs, it seems that it's the same shopping mall but with a sales representative who knows these shops, all these stores, and these sales representatives answer you which shop to use, which link to open, and so on.

So in a nutshell, I would say this is the difference between ACO and GEO David Quaid SEO: So Vlad I just want to So fir first of all it's great to have you on the show Welcome here And also um I I applaud you for for doing this in in English which is not your first language I wouldn't be able to do this in another language and I learned three of them uh when I grew up Uh so I I I also speak fast so if if I do speak too quickly um please let me know I apologize Um I I I mean English speakers think I speak fast But um you explain very well like for example how your son um can say to you Dad look I don't understand this Can you ask ChatGPT But what what you're doing is explaining the difference that Geo That sorry that LLMs and Google are different I think we know that right So we're talking to like a rather technical audience So I think the what it comes down to is like what what is the actual difference Not that this is new not that this is a shift We we know that right I I I guess the question that Edward and I would love to know and and the audience is like what makes Geo different from SEO Like Speaker: This method of marketing is so effective I had to make sure it wasn't against Google's rules before I kept using it.

It's a form of SEO I call compact keywords. Whereas most SEO focuses on putting up articles to answer questions, how, what, when, compact keywords focuses on putting up dozens of pages that sell to searchers who are actually looking to buy. These pages rank on Google and convert so much better than normal that when I discovered this years ago, I couldn't believe this was allowed. It's less work too.

The average compact keywords landing page is only four hundred fifteen words. Compact Keywords is a thirteen-hour deep course on getting sales with SEO. A customer said, "We spent nearly eighteen thousand in the last year and a half on marketing and SEO through different agencies locally, and that did nothing. We decided to take the leap on the Compact Keywords course.

We're now getting about six to eight calls per day on a good day, which is just unheard of." Another customer said, "Give it to a junior employee, have them follow it exactly as Edward's laid out. You don't have to do anything, and you're going to gain a six-figure SEO level employee just by having them go through this course." Compact Keywords is about setting up an SEO funnel that brings you sales for years and years and years.

It works with AI, it's less work than traditional SEO, and it makes way more money. You can get it now at compactkeywords.com. Back to the podcast.

Vlad: So I would say that, hmm, uh, for now with GEO, if you want to be mentioned in GEO, should do, do more job than, than you did before for SEO. What, what does it mean? So, uh, for now, like LLM, they don't rely only on data from web, from search engines. they rely also on other stuff which is not included, I believe so, which is not included in, uh, uh, traditional SEO, like, I don't know, your reviews reputation management, your mentions in PR, PR articles, uh, your mentions in Reddit and other stuff.

and moreover, for now, uh, you need to build a brand entity your, for your company. Let's say if I have a shop with, uh, running shoes, let's say that I'm a producer of the best running shoes for, to run a marathon, okay? if I want to, to be ranked in Google before, I just need to do on-page SEO, off-page SEO to buy ba- backlinks and other stuff. So but for now, if you want to be mentioned in LLMs, it seems that it's not enough.

It mean that we should get some reviews from people which say, "Okay, yeah, these, uh, these shoes are really good f- to run a marathon." We should get some mentions in Reddit as well. So it mean that we should explain to LLMs that our brand or our shoes are the best answer for, uh, for users' requests. So that's the difference.

So here you should to focus on your brand entity and your brand positioning in some case David Quaid SEO: But why why can't SEOs put stuff in Reddit Cause like I'm an SEO I put stuff in Reddit all the time Vlad: So you can do it, but you can do it also without Reddit as well. So that's, that's the difference. So, and you know, in some cases, I don't say that, you David Quaid SEO: The what we Vlad: mm-hmm David Quaid SEO: Well so I've I've been in SEO for 26 years I My last fulltime job and all the companies I work with are in SaaS I also occasionally fix legal sites and ecommerce sites and whatever Um the sites have reviews in G2 Uh we build our own review pages Um Edward has a great strategy where you buy a domain name and you add reviews to it right So like if you were Nike you could buy nike reviewscom and then you could link to all your different review profiles and you could agaggregate them Um you could also give feedback on how people can leave a detailed review I don't understand how SEOs are like um blocked or are you know I don't I don't understand where the door is that that stops SEOs from Vlad: So I would say that, you know, like here, SEO, they just help to, to be mentioned here in LLMs, so with GEO.

So but it seems that if you are ranked number one in Google search, it seems that, uh, likely you will be mentioned in LLMs, but not 100%, so. And the David Quaid SEO: Zach you Vlad: Yeah, please David Quaid SEO: You you So you you made a a a a You made a characterization of SEO that I really didn't like right Which is like you do some onpage SEO you buy some backlinks I've never bought a backlink Well I That's that's actually a lie I have bought backlinks Uh I experiment and test a lot over like over 120 domains Um I appear a lot in LLMs right And I do that via SEO I do have I actually am one of the few people with a knowledge panel I don't know if you have a knowledge panel in Google but that actually doesn't make me show in LLMs Um I can understand for example where you look at a very very big brand like a Nike or a Microsoft and they come up so much in everyday conversation But you're you're you're you're you're painting a model and then talking around it and I really wanna try to get to the heart of it right To me SEO is like surfacing any page right So I could for example have a profile on G2 and Capterra I could have a profile on Trustpilot Um I've had people uh vandalize Trustpilot so I might want to move Trustpilot down the rankings and I might wanna move G2 up the rankings right And I can control that via SEO Uh when I look at when someone asks an LLM a question it has It starts off with like its understanding Most of the time it actually has no idea where the brands are It does a search It doesn't cite Reddit It's that sometimes Reddit is returned in the results from the query fanout It's not always Google but it might be Exa it might be Brave Search but these operate a lot like Google Um it could be Bing but again that's just a Google clone And it seems that they don't know about the brand in 90 of cases until they they read this So I I don't understand how you're saying that Geo does something and can't explain what it is and that SEO is something else and it can't be this And I don't understand where that sort of like hard definition comes from Vlad: Uh, so you know, David, I think that, um, uh, here we, we should to define what is SEO is, okay?

Uh, in my, um, in my opinion, uh, in many cases in the world, because I've talked with, uh, um, many agencies as well, I'm not the, the best SEO specialist, but a little bit understand what, what is it. So it seems that in many cases, um, agency, uh, when they say we do SEO, they mean they could rank your website of your brand in top 10 in Google. So in many cases, okay? I can say that I saw a lot of agencies or people who ranked as well, I don't know, Trustpilot, G2, and all pages which you like because you usually it could be costly for some brands.

But if you define that SEO, it mean that to ranked exact website of the brand in top ranked of Google. this is… So in this case, GEO, it's a little bit different because it's not enough just put one website in the rank of Google. If you did before Reddit, I don't know, you worked on reviews as well, on reputation management, on some PR stuff, you've built a positioning for brand to deliver the message and so on and so on, I would say that you did GEO, but you called it SEO. So that's the difference David Quaid SEO: Okay but you said it must be in a it's just your exact domain I mean is that And you started off saying that's your personal uh opinion and and that's fine You know everyone has their personal opinion But I've never known SEO to be limited to the exact domain I think that's a new definition I I don't and I've never seen that in in a in a dictionary definition or And um I've been doing that for years I mean if you look at most S a lot of SEOs their X profile ranks higher Um their Instagram can rank higher A lot of a lot of SEOs don't even uh rank via SEO directly or or get business via Uh so I I don't know that that definition that you have I don't know if that's just convenient to your argument but that's not a shared definition by any stretch I don't So I'm trying to really get to like what what what is it that GEO does that isn't done in SEO Because it it it sounds like the only differentiation you have is that GEO is like SEO plus other sites but those other sites also have to rank in Google which to me is still SEO To me it doesn't mat Like a lot of people use parasitic SEO right For example like writing an article then having it appear in Medium or um what I call coranking which is um I want my G2 to rank instead of my Copilot or I want my YouTube to rank Uh which And then there's ORM which is about ranking hundreds of other pages right And ORM is a subdiscipline of SEO and always has been and like that's like 20 years old Vlad: Okay.

So, uh, David, um, sorry. Sorry, guys. So, um, know, I would say that if you are talking about as I understand in sh- in your definition, SEO as, um, uh, promotion your website in many ways, I mean, in reputation management, in PR, in social media, in Google search, and so on. say that we are talking about one thing, to, to build brand entity, to build brand authority, to, um, to promote the brand, I would say everywhere, so as a one brand entity.

So, but as I mentioned before, in my, uh, in my opinion, so in many ca- cases, so when people say about just say, "Okay, we want to be ranked in Google." So it's clear that in some, like to, to make some reviews on G2, they could help you to, to be ranked because you have a link on your website. And the same, I don't know, if you, if I sell, um, running shoes, so I could be mentioned in a listicle about the best running shoe- shoes, and it helps for, SEO and it, it also help for GEO as well.

it means, you know, like, if you're talking about the developing the brand and promoting the brand, I would say that here we are on the same page. at the same time, you know, like now you are ranked number one in Google, it doesn't mean that you will be mentioned in other LLMs because, you know, like if you look at the ChatGPT, if I'm not mistaken, they get information from Bing, so it's different, uh, dif- different search engines. Perplexity has its own, uh, David Quaid SEO: No it doesn't Vlad: Okay.

David Quaid SEO: uses Google 100 And actually um uh ChatGPT uses um SERP API to get scrape results from Google and that's why Google is suing SERP API So to say they get it from Bing is a massive overstatement Vlad: So David, am I right that your point is if I'm ranked number one in Google for some keywords, I will be mentioned in LLMs for the same keywords as well? David Quaid SEO: No So my question was really simple right The the acronym is SEO search engine optimization It doesn't say Google it doesn't say Bing it doesn't say TikTok it doesn't say YouTube It doesn't uh it it it in no way restricts anyone to anything It just says it's search engine optimization which means you do whatever you want to do to get a page to rank in a search engine So the search engine could be LinkedIn it could be TikTok it could be YouTube It it doesn't matter It's like paid advertising Paid advertising doesn't mean Google Ads right It could be Bing Ads it could be ChatGPT ads for example right If it's a payperclick model if or or it's a PP PP a CPM model So I'm just trying to figure out like it just it it the term's really simple It's like search engine optimization right Um and none of the LLMs have a search engine right They all have to outsource right So on our last Um podcast trying to discuss this If you look at um how LLMs are built right They are NVIDIA processor heavy They are RAM intensive They take a LLM which is a large language model that has a lot of variables that are basically sequences or patterns of how language is used They are not a collection of a review of every brand on the planet right Nowhere in any of the LLMs right And Perplexity isn't even an LLM Perplexity is actually a wrapper but it's still not a search engine and it doesn't have a search engine and its query fanout 100 correlates with Google right ChatGPT might use Bing and I think a lot of people believe that it uses Bing because of the relationship between them which between Microsoft and OpenAI which broke down But there is so much information about the to and fro of OpenAI using Google which means in the last 24 months if you're not aware of that you're not updating your own model of how LLMs are rag out to get content right So an LLM can understand a sentence like I want to buy running shoes I is me Running shoes are a thing And it may also associate the word Nike with running shoes because a lot of people might use Nike and running shoes say in a conversation But they actually aren't allowed to train on Reddit Reddit has sued a lot of them including Perplexity So the second aspect of how LLMs are deployed is that they are largely deployed in RAM right Because it they need access memory right Disk memory has to be retrieved from a disk and taken up into RAM and then it can be used by an app via processor You can't take a 25 exabyte which is an enormous amount of data which is like what Google stores right And put that into virtual memory and then take one million copies because you have one million users logged into ChatGPT Like there isn't enough RAM on the planet We don't actually even have a roadmap to that So the idea that LLMs can store information on every brand on the planet is further broken I was just trying to say like it's search engine optimization It doesn't in any way limit you to a domain or one page or one search engine That's all I was asking Vlad: So David, may I ask you, when you say that, um, in search engines correlates 100% LLMs, what, what do, what, what do you mean exactly?

David Quaid SEO: Okay So let's say I go to Perplexity and I'm trying to figure out for example Uh I'll give you a fun one right I want to be in the list of um top geo experts Actually no I'll give you a bit a more fun one I actually There's a lot of people who create EEAT checkers EEAT checkers right That that check to see if there's EEAT in your website And I I think this is impossible because EEAT is different to every single person right Like I might look at at at Wells Fargo's website and I think remember that time when they had all those bankers creating fake bank accounts I'm like I don't trust this company I don't care if they have like a padlock logo and a armed security guard I'm like I don't trust Wells Fargo right I don't care what they say right So up until like a month ago if you searched for um best EEAT checker whether it was in Perplexity or G ChatGPT or Gemini it would say Hey yeah you can look at all these EEAT checkers There's uh it's a great way to check if there's EEAT in your page And what we would do is then shift it to the side and have a look at the query fanout right So my prompt could be You are the Harvard professor with 185 years of experience in EEAT checkers and you want to do research in the best EEAT checker At the end of the day it's just gonna break down best EEAT checker and then it'll send it to a search engine I don't know what the queuing mechanism is right It might be ChatGPT uses its cache a lot so it doesn't incur a lot of um credits because they've got to pay SERP API for all of that scraping or Gemini which has like an unlimited live pull from Google right But it seems to be like 24 hours behind I said, I'm gonna write an, uh, a blog post about best EEAT checkers, and I ranked that in Google, and similarly, it does very well in, um, Brave Search, which is used by Claude.

And so when I look at the query fan-out, now number one for Google, which means I'm number one for Gemini and Perplexity, mostly for ChatGPT, and also for, um, Claude. And so now when you do the, um, this… When you ask the same question, it has changed its narrative, which is like, well, there's not really such a thing as an EEAT checker, but if you really want an EEAT checker, here's a list of the other EEAT checkers. And I can re-repeat that process, right? I could do that for, um, other things like best SEO podcast or SEO blogs or top SEO influencers.

I may not have control over, you know, if there's, if there's a list of like 100 people and there's only a list of 10 coming back, I might not have final control over that. But I can insert anything into that by ranking because, again, an LLM is incapable of storing 19, 20, 30 exabytes of data in its memory. It can't even store the whole of Reddit in its, in its database. Vlad: So, you know, David, I just asked Perplexity, ChatGPT, and Google, um, best EEAT checker.

So, and it sh-showed me different results in all, in LLM, in Google. So if you say that, so as I understand from your point, the results should be the same, isn't it? David Quaid SEO: No, I didn't say the result is the same. I - What I was saying, you asked me when - what did I mean that the results of the query fan-out are the same?

I'm saying that the results when, um, and Claude and Gemini show its query fan-out, those results match the search engine. I'm not talking about how the LLM arrives at the final answer. That was different. We're talking about inputs and amount of memory and amount of, of, of data that an LLM can store.

it - we're, we're establishing the, we're still establishing the difference between GEO and SEO, and I've just said that SEO is about ranking a page in any search engine and ranking any page. It doesn't have to be your domain. That, that was just a of SEO that you hold that I don't think matches the general definition of SEO for ev- for everyone else Vlad: So why do you think that Perplexity, ChatGPT, and Gemini for this - uh, and Google for the same re-re - for the same request, they have different results?

What's the main David Quaid SEO: So Google's just, Google's just going to return the pages from its index. Um, and ChatGPT and Perplexity are actually gonna take the prompt that you gave it and change it, get results that are slightly different rank order, and then they're going to synthesize it, right? And that might be based on your personal history in ChatGPT or in, uh, Perplexity. Uh, but the synthesization process is different for all of the LLMs.

That's how they're built. So I don't know why or how LLMs do synthesis or why they get a d- why you get a different result between Perplexity and Gemini. But I didn't say that they will match to Google. All I said is that the results from Google, Brave Search, and from Bing tend to correlate, or that the query fan-out returned by Google for a query tend to match the same documents that Perplexity is using to build its synthesis Vlad: I think that here is the difference when you say that it correlates, it doesn't mean that GEO equals SEO.

It just correlates. And this is a question, David Quaid SEO: C Vlad: a percentage of correlation David Quaid SEO: No, that's not what I was saying. What, what I was saying if you take a look at the prompt and you look at the query fan-out that the LLM sends to Google, and you put the same query into Google, the results for Perplexity correlate with the results from Google. I didn't - I'm not talking… As in it shows that Perplexity is using Google's results.

Otherwise, if they would correlate with Bing, you could say it uses Bing. That, that's, that's just an undeniable fact, right? Perplexity isn't a search engine. It doesn't have a search database.

It's not even an LLM. It's a wrapper above an LLM Vlad: Uh, yeah, that's true. But, you know, like, when we are talking about, uh, when you say that, uh, GER equals ACR and at the same time, you know, like for the same, for the same request, we have different results and you say, "Okay, we have like prompting not, uh, requests, uh, in LLM," so it means that at the end of the day it's different David Quaid SEO: No, I'm talking about the input documents from the query fan-out. Do you know what the query fan-out is?

Vlad: Yeah David Quaid SEO: Well, okay, so what does the query fan out to you? Because we seem to have a different opinion Vlad: So actually, you mean that if you have a request, LLM divides this request into, mm, different, I don't know, ca- ca- categories or, entities and provides the answer David Quaid SEO: I know So a query fan-out is basically what the RAG process is. So you put in a prompt, it takes the prompt and breaks it into a, a set of queries. It could be one query.

Based on the results of that query, it may have two, three, six more queries. It then sends each of those queries to a search engine. ChatGPT could use Bing or Google largely. Um, Claude uses Brave Search.

Perplexity and Gemini use Google. They send the queries Google, Bing, Exa return 10 results per query. They take those results and synthesize the final answer from that Vlad: Okay. Let's, David, move to, to the another point, okay?

To discuss another point. David Quaid SEO: But wait, are, are you saying that that doesn't happen? Vlad: Um, doesn't, uh, happen what? That David Quaid SEO: So no, I, I was just - You, you said - I, I was asking if you knew what a query fan-out is, and I just explained what the query fan-out is.

Are you saying that the query fan-out isn't true? Vlad: Yeah, it's true David Quaid SEO: Okay. So the LLM has to go to a search engine to get those 10, 30, 60, 90 results from which it builds its synthesis? Vlad: So I would say yes, but the final decision what to show is, uh, is on LLM sites.

So, and you David Quaid SEO: Okay. Vlad: here you David Quaid SEO: But it's not on geo side, right? Vlad: Uh, I would say that what answer, so GEO helps to LLMs to, to provide the right answer. So it's a different David Quaid SEO: does Geo help an LLM to provide an answer?

Is, does Geo insert code into the LLM Vlad: SEO also is not included in the code of Google, you know? David Quaid SEO: I didn't say it was. Vlad: So that's David Quaid SEO: asking is how, how is Geo helping, how is Geo helping an LLM provide the answer if it's not coded into It Vlad: So actually, let me explain. Uh, if I create a with domain name thebestrunningshoesformarathon.

com uh, built a lot of links, I don't know, uh, and put this link in top of Google for the same request, is it doable? For now in Google. Is it doable? David Quaid SEO: depends on the… Vlad: I David Quaid SEO: You can do it.

Will it work? I don't know Vlad: actually, but it's possible doable because I saw at least, I, don't know, dozens examples when people create a simple websites for the keyword and this keyword is inserted in domain name. Did you see such kind of examples before? David Quaid SEO: Are you talking about listicles?

Vlad: Um, any website for one keyword, just one website, one keyword in, it's ranked in Google David Quaid SEO: Okay Vlad: So in some cases, it's doable. Like, you can do it. You can buy some links for this website, and it will be ranked in Google high. High, I don't David Quaid SEO: Why do you have to buy links?

Vlad: So, David Quaid SEO: Okay Vlad: but if I want to, for LLMs to prove that my shoes are the best for, to run a marathon, it's not enough. So you have to build brand entity, to build trust and many other stuff. In this case, you will be mentioned in LLM, you know? So, and David Quaid SEO: Okay, so how, so what is it that you're doing that builds this brand entity that's, that's different is what I'm asking Vlad: Uh, different from what?

David Quaid SEO: You said, uh, you said that SEO gets you into the final list, GEO is what gets you cited. So what is it that GEO is doing differently? Vlad: So actually, as I mentioned before, if you're talking about like your approach when you promote Reddit, when you promote, I don't know, reviews and so on, you do a lot of stuff in terms of the SEO. It seemed that it's similar.

but at the end of the day, so I would say that one thing is missing. So when you want to push your keywords or pages in top, in top of Google, so you just focus on these pages and you don't think, for example, I think, so you don't, shouldn't rely, uh, you shouldn't think a lot of, I would say, positioning of your brand. That, I don't know, that your, uh, that your shoes will be the best for to, to run a marathon, that everyone knows about it. David Quaid SEO: why Vlad: yeah, David Quaid SEO: wouldn't you?

Why, why, why, but if you're building a webpage and you're getting people to the webpage, why wouldn't you build positioning for your brand? I don't understand Vlad: so if you - it's not obligation to do it, you know? David Quaid SEO: Like, are you saying if I go to Nike's website, it doesn't have any positioning, or if I go to Apple or Microsoft, there's no positioning of their brand? Vlad: I would say for Nike, so for different, I don't know, they have definitely, but not all of them has.

And you know, like if you David Quaid SEO: Not all of them. So ha- have you seen, so you've, you've basically started at Google and gone from A to Z, and you've gone through every website, and you're saying that the SEO ones don't, they don't describe their brand or positioning? Vlad: So let me explain the difference. So, uh, there are a lot of brands, for example, for the running shoes.

So, and if I ask LLMs about what's the best running shoes for, for my case, so usually they show me the short list of, I don't know, Nike, Hoka, ASICS, and others. So if I But to, to provide this information to LLMs, I need to, to show the case studies, I don't know, that my running shoes are the best for your case, to, to get some reviews, uh, reviews from users and, you know, like, to build a brand entity, uh, around my po- positioning and, uh, about around my positioning. So for Google ranking, it seems that it's mm, not necessary to do the same, be in top of the Google David Quaid SEO: But so you're talking about what it takes to rank a page, right?

But Vlad: Yeah David Quaid SEO: owning a brand in Google means also controlling reviews. Like I said, one of Edward's great strategies is take your domain name and have reviews.com. And then one of my strategies is to have pricing and review pages to link to those pages so you can return who's got the best profile on you.

but I just went to Perplexity, right? Um, and, and I'm just using Perplexity 'cause I'm lazy because it shows the Q of… Can I share screen, Edward? Quickly. I said, "Show me the best running shoes for a marathon," right?

And it says, "Completed two steps. the web. Best running shoes for marathon." And then it's got these results: the best running shoes, Runner's World, the Run Tester, seven best.

There's a YouTube video. There's actually, funny enough, there's no Reddit. And there's Nike, right? 'Cause they have an article called "The Best Shoes for Running a Marathon" by Nike.

So here you've got Nike, and that's from Return From Run Repeat. You've got Nike Vaporfly, um, HOKA from Runner's World. interestingly, Nike is not actually even cited a lot. I think that's just because they show up, like, as number 10.

So let's take this a lit- well, I'll just type it in myself Edward Sturm (2): So what are you doing, David? Just to narrate it for people who are listening and not watching David Quaid SEO: sorry, I'm just typing in best running shoes for marathon because that was the same query fan out. Uh, so I've got Reddit, best running shoes for marathon, Soloreview, The Run Testers, Runner's World. I've got the video, Adidas, um, Runfair, Slowtech Forum, and there's Nike at number 10, and there's Run Repeat.

So it's like the same websites, slightly different order, um, but it looks like it these based on Run Repeat, which was the third page returned. And I imagine that both of these also have in them Vlad: So David, as I see that, uh, Perplexity has dozens of different links, but not all of them were in top 10 of Google even. So I David Quaid SEO: So Vlad: know the David Quaid SEO: I, Vlad: I David Quaid SEO: I th- Vlad: other links. So, and, you know, like, it seemed that were captured not in top of Google, you know?

David Quaid SEO: So all of the links I saw were in Google, just a slightly different rank order. Nike was actually on, position 10, but on the next page Vlad: we can check, double-check if you want, but as I saw David Quaid SEO: Okay Vlad: all of the links were mentioned, uh, on the first two pages, you know? So, and that's the difference. Your You could be mentioned in, um, I don't know, in article on 10th page, this article could lead you to perple- perplexity answer David Quaid SEO: Right, but Perplexity is not a wrapper and it's not a search engine, right?

It's getting its results from somewhere else Vlad: Uh, you know, like, it's not a search engine, but, you know, like, uh, in the beginning of this David Quaid SEO: What, what I mean by it's… By the way, just, just so we get into I mean by it's not a search engine, it doesn't crawl the whole web, it doesn't index every file, it doesn't process it, and it doesn't research every brand. That's, that's literally what I mean. It built that answer right now based on those 10 results that were given.

Granted, the rank order in my Google and per- the 10 results sent back to Perplexity were slightly different, but 70% were largely the same. As to why Reddit was in mine and Reddit wasn't in Perplexity's, I don't know. that that we read from Perplexity was created based on those 10 results. It wasn't based on its memory from last week Vlad: So I can't agree here, you know.

Um, David Quaid SEO: Okay Vlad: the beginning of this year, we've made a test. We asked our technical team to analyze our server logs, so for two weeks, uh, to, to see, I would say, real behavior of Google and AI crawlers. And you know, like, what we found um, actively crawling, Google Bot, mobile and, and desktop, OpenAI, GPT Bot, we saw Claude Bot, Perplexity Bot. So these were bots from AI, not from Google or other search David Quaid SEO: Hmm?

Vlad: And I was surprised that on some days, AI bots matched Bot in, in volume. So that's why I think that it's just a matter of time, and it seems here I can't agree when you say that, LLMs like OpenAI, Claude Bot, and Perplexity Bot, they don't crawl the website. They do David Quaid SEO: What, what I'm saying is when, when it gets those 10 results, right, it has to fetch those 10 files, right? Like the file is on a server, it has to fetch it to synthesize.

But that answer is made up right now in real time, right? It's not using data that it learned. It took those 10 files right now and built that result. So on those servers, those servers are going to show a crawler from Perplexity fetching those 10 files right now, live in the mean- live in the real time.

Vlad: So you David Quaid SEO: That's why you see crawler, uh, activity going up. one of the challenges I have, for example, with, with web engineers is they al- they always conflate crawling with indexing outcomes, right? Like technically Google only has to crawl and index a file once. It doesn't have to - It keeps crawling it, right?

To see if the file has been changed or updated, but it doesn't have to… Like the more crawling doesn't mean like you're more indexed, right? So Google doesn't have to up the amount of times it crawls the same file Vlad: Shaan David Quaid SEO: So that answer is not, if, if, if, if that answer is coming from memory and from having researched all those brands, why is it doing an RAG out to, with a query fan out to a search engine, taking the files and fetching them in real time?

Vlad: Uh, so David, may I ask you the question? So, um, how do you see, how do you think, uh, does Claude, ChatGPT, uh, Perplexity, do they want, to u- to use search engines from other brands like from Google, from Bing? Or if they have its own, they use own search engine. What do you think about it?

David Quaid SEO: They don't have their own search engine. Vlad: They don't David Quaid SEO: Perplexity doesn't even have its own LLM. Vlad: if they have, if they David Quaid SEO: But that's, that's a different question. But, but now you're asking - now you're putting yourself in the position of like, are - if you were the CEO of Perplexity or Claude, would you have your own search engine?

That's… Maybe, may- maybe one day you might become the CEO of those companies, and that'll be a lovely question and a lovely interview. But what we are here today, right, is I just put in best running shoes. It went and fetched 10 files from… It, well, it got a set of results from a search engine with the results. It went and retrieved the 10 files, and it synthesized an answer.

It didn't build it from looking at a value proposition from last week, right? It built it right now in real time. The crawler went and crawled 10 different websites and received - and retrieved 10 different pages and built a synthesis right here in front of us from just those 10 Vlad: but the point is that it were, uh, so, uh, these 10 links, uh, were getting not from the top of Google or Bing or others. So LLM decided on its sides what to David Quaid SEO: Where did it get it from then?

Vlad: No, no, no David Quaid SEO: Where did it get those 10 pages from? Vlad: So decides which pages, uh, which links David Quaid SEO: but from, from, from where? Vlad: from the web, not David Quaid SEO: So it just went to the web. Vlad: not from the ranking David Quaid SEO: went to the web and just, and just said to the web, "Give me 10 pages about for running shoes with marathons."

And then the same 10 pages matched exactly the same 10 pages in Google. Well, 90% of the pages in Google search results Vlad: So, but you know, they, they get these pages not from the top 10 or top 20, you know? So they have a memory, definitely. They s- they have a memory of David Quaid SEO: state Vlad: from search engines.

That's true. David Quaid SEO: They have a memory of the whole internet. So we're going back to what I said at the start. Google's memory of the internet and Bing's memory of the internet is around 25 exabytes.

Uh, an LLM manages about 100 gigabytes. So there's a, there's a, a, an order of magnitude of about 99.999% missing here, right? Vlad: yeah David Quaid SEO: None of, none of the LLMs have a copy of the World Wide Web, right?

They are - No, no one else is debating that LLMs don't have a copy of the web, and they go to… Right? For example, it - Bing is suing - Sorry, Google is suing SERP API in the US courts for scraping its results and giving them to OpenAI without paying for them. It's a, it's a literal lawsuit that's happening right now. Why would SERP API be scraping Google results and giving it to OpenAI at the risk of being sued by a trillion-dollar company if OpenAI has its own copy of the worldwide database, which we've already established can't fit in 100 gigabytes of RAM?

Vlad: So David, uh, let's look at the SEO on a different angle. So when your clients David Quaid SEO: No, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no. It's really simple. It's a simple, it's a simple question of mathematics, right?

Vlad: Okay. David Quaid SEO: An LLM runs in under 100 gigs of RAM, maybe even 20, 30 gigs of RAM. many gigabytes of RAM is in 25 exabytes? Edward Sturm (2): But David is doing the math right now David Quaid SEO: I am asking.

Okay, so it is 25 trillion megabytes. So you're saying that ChatGPT has gone out to the whole web, it's researched every brand, it's gone, "Oh, wow, I love that Vlad wrote a fantastic story about the Top Bar"? Vlad: No, it's not. David Quaid SEO: It's not So what's it doing?

Vlad: Um, so, um, so my point is that, you know, like, uh, may I ask you, when the clients came to you, you, you mentioned that you're like 26 years in, in the SEO, and when clients came to you and say, "Okay, we want to be, we want to do SEO," so which, um, numbers you measure when you do SEO to, to track your result? David Quaid SEO: I'm, I'm sorry, yeah, w- my clients have nothing to do with this. Let's just come back to the simple problem, right? Google has a copy of the World Wide Web.

It is about 19 to 25 exabytes, right? Uh, that is about 25 trillion gigabytes, sorry, megabytes of RAM. You're saying that an LLM didn't have to go to a search engine to get those results. It went to the web.

I'm asking you to reconcile that enormous difference between what it actually takes to store the World Wide Web, because you don't know what's in a webpage without fetching it, right? So I think people, like, have this magical idea that the World Wide Web is out there and it's available in a telephone directory, and you can just ring it up and go like, "Hey, give me this, give me that, give me that." The only way you know what's in a webpage is you have to have a crawler which fetches the webpage, an entire copy of it, it down, extracts the text of it, and it's indexed somehow.

So you don't know if Nike have their value proposition in this page or this page or across this page, and like you said, it could be in Trustpilot, it could be in G2, it could be in Capterra, it could be in Reddit. So the only way to know where Nike is in all of these pages, you have to get a copy of all these pages, and that means you have to store it. To store that is about 19 to 25 exabytes. An LLM has to run in under 100 gigabytes of memory.

Where is it storing all of these pages so that it can retrieve it to get you that answer? Because this is one example of best marathon shoes. If I say, what is the best airline? What are the best speakers?

What are the best AirPods? What are the best cars? What are the best roads? What are the best bridges?

What's the best pasta? What's the best bolognese? All of that has to be stored. It can't just magically make up an answer in demand.

So in order to store an answer for every single one of those questions, you need 25 trillion gigabytes of memory. It doesn't have 25 trillion gigabytes of memory. So what - in what unicorn universe is it storing this data somewhere? Vlad: So that LLMs, they should use, uh, Google, Bing, and other search engines.

They don't have … They have to use it, you know? But my point David Quaid SEO: Okay Vlad: that, you know, you showed, before you showed two screens, and when you type the same request in Google, so Reddit was on the first place. in this case, if you want to promote your brand, uh, for this request in Google, uh, so the good way to be published on this Reddit post, because in this case, your brand will be mentioned there. But at the same time, when you do the same in Perplexity, we saw dozens of links, but we didn't, David Quaid SEO: No, we saw 10 links and, and Stop.

We saw 10 links, right? It, it says 10 links, right? It wasn't dozens, okay? Vlad: okay, okay David Quaid SEO: And all- most of those links correlated the difference in The difference is I'm logged into Perplexity and I'm logged into Google, right?

I wasn't trying to show a one-for-one likeness between Google and Perplexity. That wasn't the aim of the game. The aim of the game was to demonstrate that my prompt is broken up into a query. That query goes to a search engine.

The search engine uses its algorithm to return the results. Perplexity can't tell Google how to change the results. ChatGPT can't tell Bing or Google or whatever it's using to change the results. It doesn't have access to its database.

It has to go to a Google search bar and put in a query. That process is called the query fanout. it is still using Google's algorithm to return the pages. I understand that there was a difference of maybe 20 to 30%.

We can account for that however, right? It's not like it was a completely different set, and it also wasn't that it was somehow pre-trained, right? Vlad: So as you mentioned that there is a difference like 25 or 30%, and that's the difference between Perplexity and Google Answer. So it means that it's David Quaid SEO: How do you mean it… Ho-hold on a second.

How is that the, the difference? W- Perplexity, h- so how does Perplexity get a different answer to Google then if it's using Google? Vlad: we just show, showed it in, in your request on your screen David Quaid SEO: we showed a, we showed Google giving a slightly different number, uh, or slight- a slightly different rank order of the pages, right? Vlad: But it's different David Quaid SEO: which is… Okay, but so what, what makes the difference?

Vlad: So in this case, like Perplexity decides what to show, you know, look, David Quaid SEO: How? Vlad: based, David Quaid SEO: It's only given 10 results back. How… But it, it's given 10 results back. How does it go and decide?

Where is it, where is it now getting these other pages from? Vlad: So that's a question David Quaid SEO: Yes Vlad: So as I … So you've, you've said that, you know, like not all the, like there is a difference as you, as you said before, there is a difference between, you know, David Quaid SEO: I said there was a difference in the rank order Vlad: Yes. and that's the point. So if you work on LLMs, so in some cases you don't, you don't need to be mentioned in the top of the Google.

So you, you could be mentioned on the 10th page of the Google, and in this case, it helps you to be mentioned in LLMs. But at the same time, in Google, you, you can, if you want that a user click on you and visit your website and know about you, there is just like 10, top 10, top 20, top 30 David Quaid SEO: Yeah, it's, I d- the question was how does Chat- how does Perplexity change Google's Vlad: So it's, um, Perplexity has its internal weights, internal algorithm how to show it.

So, and it's David Quaid SEO: but, but okay, but where is it getting its pages? Where is it getting its index of pages to apply this weight to, and why is it 90% similar to Google's? Vlad: So, you know, like if you say that it's 90% similar to Goo- to Google, it means that it's 10% different from Google. So, and it means it's, you know, like David Quaid SEO: Okay, but I, I s- but where - if, if it's not using Google, where, what is it using?

Where is it getting all these other pages, right? Like I just said, in order to pick pages, you have to have a copy of every page on the web. Vlad: Yeah David Quaid SEO: It - we've just shown that it doesn't have the memory storage to store the whole World Wide Web Vlad: So actually it uses Google. So I don't say that, you know, like, that, uh, when I say that every LLM, so users use, use web to get the, the information, it's clear that they use, like, Google, they use Bing and other engines.

They don't, they don't have… They have to do it. So but my point is that the answer in Perplexity and Google could be different. And you know that you've mentioned before, like, RAG approach, so and I think that the augmentation, so the second and the third step in LLMs, they provide a, in this case, they provide a unique instead of the Google, because they can influence only on the first step, retrieve David Quaid SEO: But how does it influence it? Like, you, like, are you saying it has access to Google's index?

Vlad: Influence what? Sorry David Quaid SEO: You just said it picks, it has its own weighting, so it picks different pages. The h- where is it finding those pages to pick them from is what I was asking Vlad: they found these pages from, from search engines definitely David Quaid SEO: Okay, so in - all, all I'm saying is my Google is re- is returning personalized results. So I'm not - I wasn't trying to establish that the query fan-out is exactly the same, right?

I can, I can get them exactly the same if, if, if I need to. I can log out of everything. I just don't have the time to do it. But you're saying that the LLM uses its own weights and preferences.

I'm just trying to establish if it's, if it's using its own weights and it's different to Google, where, where is it getting it other pages to apply those weights to? How is it doing that? That's all I'm asking. Vlad: So at the end of the day, LLM, they have its own mechanism to show the answer.

So they definitely, uh, have a - they definitely got all the information from search engines, but finally, they decide what to show, and that's the difference. you know, like, we can use different requests in different search engines, in different LLMs for the same machine, PC, MacBook, doesn't matter. So, but we will see different results. David Quaid SEO: So then tell me how GEO is different from SEO that it can s- influence that final step Vlad: So the difference is that LLMs, they decide which brand to show and which company to show.

So this is an additional layer which doesn't exist in search engines David Quaid SEO: It doesn't exist in search engines. So how, how is that created? How - So you're saying that they do this for every product and every brand Vlad: So actually for every request, they gather the information David Quaid SEO: Okay, so in order to have an answer for every request, they must have information on every brand, right? Vlad: So they get this information from search engines.

So yeah David Quaid SEO: Okay. But how is it different from SEO? Vlad: So the difference is that they, LLMs and its own algorithm, and weights, they decide what to show. So that's the difference.

So, David Quaid SEO: So if Google says you should use, here's Nike, uh, how does the LLM say, "Well, I don't want to use Nike"? Vlad: So because, um, they have these algorithms and this approach, it's like, uh, included. They don't show it how, how they decide which brands to show, but from different searches from different companies, I don't know, from, Profound, Evernote. So they've done, like, thousands of different requests, and they understand that in this case, you should want that visibility of your brand in LLMs li- um, will be higher, so you need to do the different steps.

But … And usually all these steps leads to brand entity, your positioning, work with the reputation management, and SEO David Quaid SEO: So where, where does this brand entity live? Vlad: So actually brand entity live in the web in the search engines as well. David Quaid SEO: Just out in the web. Vlad: Yeah David Quaid SEO: So how does, how does… so how does ChatGPT or Perplexity go and find this brand entity?

Vlad: Uh, so they get the information from search engines David Quaid SEO: Okay, so this idea of them all browsing and browsing more and building their own index, that's gone now. So now they're getting it from the search engine, but the search engine can only return X amount of results, and all of the results it returns are based on its algorithm. So where is the LLM getting this difference of opinion from? Vlad: So it's their own algorithm, so we don't know, like, what's inside LLMs and how they decide, and I think that nobody knows.

But David Quaid SEO: Okay, so nobody knows. So does, does that mean, does that mean Gio doesn't know? Vlad: So I mean that we can guess what influence on, uh, GEO the same as for SEO. David Quaid SEO: can't SEOs, why can't SEOs - But why can't SEOs also guess?

Vlad: Sorry, David. Nobody knows the, um, the formula of Google, you know, like, but we are guessing what can help for a brand for SEO to be ranking in top of Google David Quaid SEO: I'm pretty sure we do know the formula because it's a patent and it's published, and you can actually chat to ChatGPT about it all day. But if you come back to nobody knows how the LLM decides the final 10, right? how does geo know more than SEO if nobody knows?

Vlad: I'm, I meant that, uh, you know, like we don't know the exact formula for different David Quaid SEO: Okay Vlad: what they use. But at the same time, we have like different researchers, we have, different and we can guess what helps you to increase visibility. And when you start David Quaid SEO: And so what I'm asking is Vlad: your visibility is increasing. That's the point David Quaid SEO: So I'm asking what stops an SEO from making the same guesses or the same research or the same experiments or the same Vlad: Uh, you know, like, you can do the same for SEO, but at the end of the day, it's, like, different platforms and you can, cannot measure, I don't know, if you want to see the traction or trends for your AI visibility.

So at least we should use different tools. You know, like, we don't have, David Quaid SEO: So if I go to Perplexity and I say, "I'm trying to work out why I'm not recommended in the top 10 SEO YouTube podcasts," it's gonna say, "Look, stop. You're an SEO. I can't tell you.

You have to come back as a geo, then I can tell you." Vlad: It doesn't work in this case. So, you know, like, David Quaid SEO: It doesn't? Okay.

So I'm asking you what stops an SEO 'Cause you just said the data comes from a search engine, right? So if the search engine's giving the results and SEO is about controlling the results that a search engine gives, what is stopping an SEO from arriving at the same outcome as a GEO? I, I, that's my question. Vlad: So, you know, like, what stops?

Um, let me show you an example. I've, um, uh, maybe one month ago, uh, I also have a marketing agency focused on crypto projects, and one of the crypto tax platform, uh, they had a problem with, uh, organic traffic because it's dropped. So, but at the same time, when we made an AI visibility audit, we show the difference trend. So it's, it was growing, so it was uptrend.

and in this case, David, what do you think? what, what the reason why it's like different trend for one brand in search engine David Quaid SEO: you asking me to Vlad: in LLMs? David Quaid SEO: Are you asking me to comment on a domain that I've never seen before? Vlad: Okay.

Um, okay, I don't have this website opened right now in Ahrefs, but, but okay, uh, it's, it's good. But at the same time, let's, let me say that in some cases, so your website could be ranked on Google, but at the same time not ranked on, mm, LLM and vice versa. So sometimes it happens. It may be you've seen such examples before, I don't know.

that's David Quaid SEO: So what I find is that the prompt doesn't match the query. So when I find what the query is and I make the website rank for that query, then the website is quoted in that LLM Vlad: So, but if the request was the same, the exact like in, uh, search and exact in LLM, so you also think that it's, these are like David Quaid SEO: How can it be? I - we don't know - we don't have control over what users put into ChatGPT or Perplexity, right? We can see that there are thousands and thousands of citations, right?

You don't know what people put in any more than I do, right? Vlad: Hmm. Edward Sturm (2): Vlad, what - Vlad, what tactics do you do at Stive to get clients more AI visibility? Vlad: Um, so let me, let me comment, uh, David, uh, when David mentioned that we don't know what people prompt in, uh, ChatGPT and other LLMs.

And, um, I would say that true, so because we, we don't know what kind of prompts people use. But, um, I would say that here we need to be… Here we can find, uh, some, uh, possible prompts and, uh, requests from SEO, from, from like Google Search. But if you're talking about what to do to be, to be mentioned in LLMs, I would say that firstly to, uh, to make an AI visibility audit, usually you do like, um, of requests during, I don't know, one week, two weeks, like 100 requests per day for different keywords to understand where are you right now and where are your competitors as well.

So then to understand what do you want to get and what kind of, what brand entity you want to, to have and to that LLMs knows, LLMs know about you. So, and start to work on this entity and gaps which you see between you and competitors. So all David Quaid SEO: Where do you get that list of keywords from? Vlad: So actually, this keyword list we get from paid ads from Google Search Console, and then we add, uh, possible prompts based on their, behavior from, let's say, based on their guessing what kind of request people can ask.

So, and then we create a segment of different keywords to use in LLMs to make an ana- analysis David Quaid SEO: So you get a list of keywords from Google Ads, Which is based on Google Search. You then run, create prompts based on that, what you guess the user's prompting, and then you look to see if the website is visible in the returned answers. Is that right? Vlad: Yeah, as we know, David Quaid SEO: And but you don't know the volume of how many searches there are, right?

Vlad: So that's David Quaid SEO: So it could be 10 for this one, could be 100 for this one, could be zero for 20 of them, right? Vlad: So we don't know the volu- volume in, uh, LLMs because they don't provide this information. but at the same time, you can check it in your Google Analytics, I mean, the traffic from LLMs, at least the whole traffic from LLMs. And usually we can guess that, uh, you have, let's, let's say 1K, mm, clicks from LLMs per month, usually it means that you have, like, four times more of your brand without a click.

yeah David Quaid SEO: And you know it's four times because I looked at, um, one of my sites or one of my projects that has nine hundred and forty million citations in the last - since May '15. I look at the amount of traffic coming from LLMs in GA4, which you also have to discount 50% because people opt out of cookies. Uh, the diff- delta was somewhere around where we got two thousand clicks from nine hundred million citations. it's like a lot less than one in four.

It's like one in - It's like back to the exabyte versus megabyte data. But I don't understand because I do the same thing as an SEO. I don't understand how GEO does this differently, right? I can go to Google and Semrush and Bing get a list of keywords.

I can get a tool like Profound, put those tools in there, run a lot of prompts. It'll tell me, "Here's where I'm visible, here's where I'm not visible." You don't know what the volume is. I don't know what the volume is, but I can then see where I'm not visible.

Then I can take the prompt, find the query fanout, modify the page or add a new page, then I will notice from all of my prompts that my visibility percentage increases. I don't understand how SEO and GEO is different Vlad: So, uh, it's different because you have to measure at least two different things. In SEO, you measure like clicks, but as you mentioned, like in David Quaid SEO: Oh, I, I'm sorry. Vlad: Yeah.

David Quaid SEO: I'm sorry. I had to measure one thing in SEO Vlad: But, uh, how can you measure, I don't know, visibility in I- uh, in SEO? What do you mean here exactly? David Quaid SEO: So the first thing I measure is what is my visibility for the number of - for the different searches, which is exactly the same process as meas-measuring what is my visibility for a prompt, right?

Like, in other words, I go to a search engine, I put in the query, I go to an LLM, and I put in the prompt. If I'm not in the results in either, I'm missing, right? Is, uh, what's, what do you do that's different? Vlad: So different that you can be, you can be visible, you c- you can have like, I would say that it's similar, you know, but the platforms are different.

I mean that Perplexity, David Quaid SEO: platforms are different? Vlad: ch- and Cha- ChatGPT, different platforms. That's why there are like additional David Quaid SEO: but Profound, Pr- Vlad: Yeah David Quaid SEO: Profound will run those in all of those LLMs, right? Or Peak or, um, Brandy or whatever, right?

It'll, it'll take the same prompt and put it into all of the LLMs, right? So Vlad: Yeah David Quaid SEO: getting the same data as I am, right? Vlad: Um, yes David Quaid SEO: Okay. So you don't know the same things I don't know, and I don't know the same things you don't know, I can do the same things you can do, and you can do the same things I can do Vlad: So so David, I would say that, you know, like when you, if you go to Profound and to try to create a AI visibility audit for the same brand, and if we get all the possible requests we have in mind from our business experience, I would say that we get the same results.

I mean, if you go in Profound in terms of the, our AI visibility score, possible David Quaid SEO: Are you talking about the difference between your tool and Profound? Because I'm not talking about the difference between… But you can tell us about your tool if you want, but I was just talking about what's the difference between SEO and GEO, that's all Vlad: Um, I'm talking about that, you know, like if can, we can measure, uh, your AI visibility presence in different LLMs, uh, with David Quaid SEO: like Profound or Peak or Brandy or Vlad: Yeah.

So, and clearly understand that we can rely on these figures here and now. So, and it means that if you work on your AI visibility, so you can see that trend, is it increasing or decreasing, and to work on it David Quaid SEO: Okay. So what's - why - don't understand. Does - Are SEOs not allowed to do that or Vlad: So, you know, like you can do, uh, you can do the same for SEO.

I don't see a problem here, David Quaid SEO: Okay Vlad: but as I mentioned you before, uh, so, uh, I'm not sure that everyone in the market when they say that they work on, on SEO, they, uh, they work on reviews, reputation management, brand entity, and many other stuff David Quaid SEO: I can tell you, I can tell you right now that when s- one person says, "I work in SEO," and another person says, "I work in SEO," they could be doing two completely different things because SEO is not a license, right?

It's not a prescription. It just says search engine optimization. You do things to optimize things for search engines, right? So if someone is doing like TikTok that is nothing like Google SEO.

It's like it's not even s- the same, right? I, I can't, I can't rank a video in TikTok if it was made by, um, Tom Cruise and Brad Pitt. Why? I, I suck at it, right?

but I can rank a Fiverr video on YouTube in like six seconds, right? uh, granted, uh, S- different SEOs do different things, and because there's no license, there's no qualification, there's no definition of what SEO is. It - SEO is a description of an industry, right? All I'm saying is why can't an SEO - 'cause I, I do this all the time.

I go and I take keywords from Profound. Maybe I don't assume that for every four clicks I get to my website, there's four citations. I know that there's actually tens of thousands more. I can go to Profound or a different tool, or I can build my own.

Also, um, Ryan Jones on X has a free tool called SERP Recon that you just have to bring your own API key, and it does the same thing, right? You're guessing prompts. I'm guessing prompts because like you said, nobody knows what the volume of prompts is. I'm just asking why is it different for an SEO and a GEO because it sounds like you and I do the same thing.

Vlad: So, mm, you know, like if you, if I, if I work on, uh, let's, let's come back, I don't know, three years ago when we, we don't discuss, uh, LLMs a lot. So, and if I do the same, like I work on, on my brand, on brand entity, on reputation management, I don't know, on PR, on social media, on Reddit, on website SEO, and like three years ago, if you say that it's SEO, now it's GEO. So, but the difference is that before you… I'm not sure that every company and every brand, uh, worked on Reddit a lot, on, uh, social media, on reputation management just for ranking on Google David Quaid SEO: So I started in - Like, I was born time ago, right?

So I was in university in 1995. I dropped out of university, and I started building websites, installing internet. So back then, the internet, we installed over modems, and they made that, like, ting, ping, ping sound, right? Does that mean that that's not the internet because we used to use modems and now we don't have, like, modems or the sound, or is it still the internet?

It's still the internet, right? Back then, Reddit wasn't there, and if you wanted to rank in Yahoo or if you wanted to rank in Mr. Jeeves, you would put the same keyword in your page title multiple times, and that's why we have the joke, years later, an SEO walks into a bar, pub, public house, drinking establishment inn, right? Because that's how SEOs used to do it.

then search engines changed, and we stopped doing that because it no longer worked. Does that mean that that's not SEO anymore? Vlad: Uh, so and now it's the new step when we, when there is a GEO, when there are a new bunch of companies and, you know, it seemed David Quaid SEO: Okay Vlad: I think that a new bunch of companies like Profound and, and others, I would say that entire category of new tools, and, David Quaid SEO: know they say it is, it, it's related to SEO, right?

Vlad: So, and you know, David, it seems that we are moving to, um, slightly moving to a, to a stage with, um, retrieval answer. So, and you know, like in this case, uh, I would say that model can from its weights without searching at all. So, and David Quaid SEO: you, you keep flip-flopping between this, right? And I'm just - I'm gonna just go back to, like, you need 25 trillion gigabytes to store the web, and none of the LLMs have that.

So let's just park that for a second, right? 'Cause that's, that's just you inserting, uh, an, an appeal to, to change. It's a logical fallacy. in 2005, when I founded my company and we were doing SEO, we used to use a tool called WebCEO because there were a thousand search engines, and it would go to each search engine.

We would take keywords. We got them from Google Ads. We took those keywords, we put them into that, it ran the query. Now you're saying it's all changed, it's a completely new platform.

You go to Google Ads, you get the keywords, you put the keywords in a tool, the tool goes to the search engine, and it measures back the brands. And I'm just wondering, where is the law that stops SEOs from doing that? Because that's exactly what I've been doing all this time, and I just don't understand how - Have I been breaking a law or am I gonna get my SEO license revoked? Or… Vlad: So, you David Quaid SEO: 'Cause I, I, I do the same thing, so Vlad: So David, you base your arguments based on the, on that, that LLM, uh, search, uh, engine every time they, they want to ask something.

But at the same time, for example, Claude, uh, so you can, uh, put checkbox, uh, web search if you want that Claude use web search in your, in your request, in your prompts. So but if I don't put it, so LLMs just gets information from their, from their, from their memory. It's clear that this memory they got from, mm, search engines, but the decision what to answer on LLMs its- itself David Quaid SEO: Yes. So where do LLMs, if you go and you, you build this sort of like brand entity, where are LLMs finding it?

Vlad: So they fi- they found in search engines definitely. They c- so but their final decision what to f- what to show, so it's, it's included in their algorithms. So they David Quaid SEO: Okay. So you, but you said that you Vlad: or Bing or David Quaid SEO: Okay.

Vlad: other search engine David Quaid SEO: so back to Edward's, back to Edward's question, what is it that you do to get this entity in front of the LLM? How do you get that LL, uh, entity for your brand so that LLMs can find it? Vlad: So actually, uh, if you know your positioning, then you have to clearly understand what's, what's your brand about. I mean, how do you, mm, describe your brand in, in, in the web on, know, on your website, on, in PR, media and so on, to prove LLMs that, to prove LLMs your brand entity and positioning David Quaid SEO: But the LLM just reads it and just goes, "I 100% believe this.

If, if Vlad says that's what his brand does, I must believe this. I just take it at face value." Vlad: it does David Quaid SEO: Same as a search engine Vlad: Um, can't agree with you. So let me show you an David Quaid SEO: Okay, 'cause we, we created, we created an example of, um, what was a funfluencer, and before we asked Google what is a funfluencer, it got nothing, LLM nothing, ChatGPT nothing, Perplexity nothing.

And I just created a one-liner, and then, uh, the following day we jumped on a podcast and we, like, asked Perplexity what's a funfluencer. It wrote like a whole page about what a funfluencer is. So like yesterday it didn't know what a funfluencer is, and now it's like, "Hey, fun, this is what a funfluencer is," right? So, um, it seems like they both take it at face value, but couldn't have got there without Google because I didn't give it to Perplexity, right?

I didn't go to Perplexity and say, "Here's a funfluencer." I didn't build a brand around a funfluencer. I didn't do any PR around a funfluencer, it got it from somewhere, right? And the only place I gave it to was Google and Bing Vlad: So um At the end of the day, as I mentioned, David, so, so they got information from search engines, it's clear.

But, uh, as I mentioned before, you know, like in Claude, you can choose to use web search or not to use web search. And if you don't use web search, this, your, your experiment, so it means that they, they will show nothing because they don't have in its memory. So that's why, you know, like, so, and, uh, not all LLMs use web search for every request David Quaid SEO: You know, not every request, right? They've got a cache.

It's expensive to use, right? But we cut - LLM - So you keep like missing the question, right? If you build your brand entity and you say like, "Well, I'm this great brand because I have these reviews and this press release," because obviously people never lie when they do press releases, right? It's where - Like especially politicians, they've never lied.

Like never has anyone used PR to, uh, confuse people or spread disinformation. It's just never happened. It's automatically… When I see a press release, I'm like, "This must be the truth." Like you can go to court, just go like, "It's in a press release, it's fact," right?

Where, where are they finding this? Are you s- You're, you're, you're back at they have to go and crawl the whole web and build a whole copy of the internet to figure out what brand is what or is it something else? Vlad: So actually they, uh, they rely on the information from source from web, but they decide which information to get and which not. And moreover, as I mentioned before, they ho- they have their own crawlers, because if David Quaid SEO: So Vlad: look at your David Quaid SEO: s- Vlad: you will see crawlers from different LLMs, then they decide David Quaid SEO: if I - Vlad: what to show.

David Quaid SEO: So what if I, what if I publish a page that says, look, let's just say, let's, let's, let's for argument's sake say we were talking about chess, right? And I said, "You know what? I think Vlad's opinion on chess is wrong." I write a blog post and I say, "Look, I've been playing chess as long as Vlad, but I've beaten him more times," or, "I've beaten other people more times, and I've won more."

I write an article, right? And the search engine gives that back and in first place, and the LLM says, "We're gonna take David's side of the story." And then imagine it like a week later, you write a rebuttal to that, and you write a blog post saying, "Look, I think David's wrong. Just because he won more games, he's older than me, and he's played more games.

But if you actually look at how many games I've played, I've won more games, so I'm better than David." And more press releases link to you, and then my website fall - my webpage falls out, then the LLM only has your article to base it on, right? So my article has gone away, and because I don't rank in Google anymore in the top 10, the LLM isn't going to use your article anymore. Sorry, my article.

So it's going to change its narrative and its synthesis, right? Or what if I delete my page from Google, then it's no longer findable, right? Then all it has is your account of the story to go on. My story disappears, right?

Vlad: So, you know, like, um, we don't know what in the memory of LLMs and maybe they have in their memories all of these pages and they can accept it. Like then David Quaid SEO: have… So now we're back to the 100, the 25 trillion gigabytes of data lives like, it's like a spare paper basket on the cloud that just magically appears. Vlad: like David Quaid SEO: It's like free, free space Vlad: Yeah. So but, you know, like, um, you mentioned that, um, your position is that, that all information, uh, which LLMs gets from the internet, they came from search engines.

But I mentioned that you can look that crawlers bot- bots, they also go to the websites get some information, and you can see it as well. So that's why they possibly can mix it David Quaid SEO: I said, but I Vlad: that they don't rely only on this information from search engines David Quaid SEO: So then the crawler, order to get of that information, has to crawl the whole World Wide Web. Yet the crawler - So if, if, for example, if I go to my Bing AI performance tab, right? If the crawler is building an index and keeping an index, why is it coming to my website every single day?

So like for example, for my SEO site, I think I got fifty-four thou-thousand citations last ninety days in Bing, of that it's only fifteen pages and I have three hundred and eighty-four pages. Why is it coming back every single day to the same eighteen pages? Or if the question isn't asked, why isn't it going to the other pages? It's just going to the pages that were in the, were, were asked.

So like for example, on Monday it says, "Who are the top SEO experts 2026?" And that is asked, I think, a hundred and twelve times. So if I look at my log, I can see, um, Copilot or ChatGPT come a hundred and twelve times to my blog. day, nobody asks about it.

They only ask about maybe EEAT or crawling or another question, but they don't ask about the top SEO experts 2026. Then there's no crawls to that page. And then the next day I see there are twenty questions, and then there are twenty crawl requests that match the twenty questions. So why every ti - there's only eighteen pages, right?

The, like the text value of that is maybe like two or three kilobytes, right? It's next to nothing. over ninety days, if the same nineteen pages are being cited, why is every single day a thousand times the crawler coming to just the same nineteen pages? Vlad: Um, mm, because they don't want to rely just on the information.

If you're talking about cro- you are talking about crawlers from David Quaid SEO: But it's coming back to my - Hey, it's, uh - I'm talking about if I go to the Bing performance report, I have 50,000 citations created from the same 19 pages. I'm happy to bring it up. Over 90 days, it's going the number of queries or the number of citations, it's going the same number of times. So if there's 20 questions, it's crawling 20 times.

If the next day there's no questions, there's no crawling. On Wednesday, now there's 100 questions, there's 100 crawlings. Why every time there's a question, it crawls again and again? It's the same 18 pages Vlad: Uh, so you mean the crawling from Bing, right?

David Quaid SEO: No, y- my page isn't in Bing. My result is in Bing. They don't get the page from Bing. Vlad: And when you, David Quaid SEO: So it's coming to my page Vlad: Yeah.

Yeah, David, when you say about crawling bot, which exactly bot you mean? David Quaid SEO: So this case, we're talking about Copilot, right? Or ChatGPT, because they use, uh, Bing for their RAG, right? So if, if there is, uh, there is 100 citations for one page, that page is fetched 100 times in one day.

So over 90 days, there's 50,000 citations for the same 19 pages, and every time there's a citation, the crawler crawls the page. Why is it crawling the page every time there is a citation? Vlad: I would say that the crawlers from LLMs, they don't want to rely just on the search engines, and if they want to get additional information to, that, or to get some David Quaid SEO: But why Vlad: your source, they just go directly to your website and crawl it David Quaid SEO: So how come they're only crawling it when it's in a citation?

Vlad: Um, so that's, you know, like, that's what's behind the scenes, I mean, in terms of the LLMs and their algorithms, how they do it David Quaid SEO: But if it's, if it's, if it's saying that it's relying on its own information, why - So where are we? We're in the 25th of June. So why in the 25th of May is it crawling the same page? Then on Monday it's crawling the page, and then on Tuesday it's crawling the page, and on Wednesday.

And if there's 10 searches, it's crawling it 10 times. it is no - if it's not in the citation, stop crawling it. So why is it crawling it so many times? Vlad: not the question for me.

I'm not a, I don't know, tech developer o- of crawlers, David Quaid SEO: you just said that it, it fetches its information on its own. It doesn't want to rely on Bing's information. Why does it keep coming back to my site to get my pages every time Vlad: so David, I've said that they got information from search engines, that's true, but at the same time they have David Quaid SEO: Mm-hmm. Vlad: own crawlers which can go to your website directly during search or, I don't know, in David Quaid SEO: But how come it's only getting those 19 - But how come it's only getting those 19 pages?

Why is it… I've got 384 pages. Why is it not getting those other 384 pages? It's only getting the pages that Bing is giving it. Why isn't it coming to fetch the other pages?

Vlad: So it's a, it's a question for developers of crawlers, but okay, uh, Edward Sturm (2): Wait, Vlad, I, Vlad, I have, I have, I have a, I have a question. Um, Vlad: The Edward Show Edward Sturm (2): you know, we - you, you asked me over email, like, how should I prepare for this? And I said, "Oh, you know what would be cool? If you had, like, a case study for - or, like, a few case studies of, of things that you did where you… Like success stories and how you actually got brands more AI visibility and the exact tactics that, that you did that moved the needle."

I think that would be really cool to share Vlad: Okay. Okay, let's do it. Um, I have a maybe, uh, not success case, but maybe negative case, but it's, it could be an option. so, uh, you know, like, uh, I have an agency in, uh, crypto and, like, la- last update of Google, uh, our company and, uh, our competitors were pessimized in Google in ranking.

Uh, but at the same time, you know, like, uh, we were growing in, uh, um, in LLM answer in AI visibility. So this is l- just one example of how it could be, how it could happen. for us, I would say that we, mm, if you measure it not in, um, mm, visibility clicks and so on, it's important part of, uh, traffic and leads uh, potential clients which they are looking for marketing agency for crypto brands, they use to find the right answer. And yeah, for now, unfortunately, we were pessimized in Google, but at the same time, sa- thank- thankful to LLMs, we are still getting leads and clients David Quaid SEO: I'm just trying to understand the difference, um, 'cause I think we do something similar Vlad: actually, uh, we have like the main, uh, in my case, uh, the main keywords for Google is like crypto marketing agency.

That's I would say the best, uh, option and maybe around 80% of all requests, hmm, comes from, uh, come from, from this group of keywords like crypto marketing agency, marketing agency in crypto, and so on. And unfortunately, I don't know why, but last update, almost all of the players in the market were pessimized. It means that from position one, we had like a position one in, uh, United States. Now, now it's lost.

I don't know, it may be, I don't know, 90 or something like this. So it's just the last update of Google. So, uh, unfortunately, but at the same time, we are still getting leads from, um, from, from LLMs. Edward Sturm (2): I wanna, uh, did you do any exact tactics?

W- like, uh, was there any like AI visibility optimization that you did? Vlad: Um, definitely yes. As I mentioned before, we have started to do like brand entity, and, um, we definitely did, I don't know, Reddit stuff and blah, blah, blah. But at the same time, you know, like, um, we found that LLMs ask to provide exact, uh, answers for exact requests.

Let's say that a user come to ChatGPT and asks, "I have a crypto exchange. Please advise me the best agency for a crypto exchange." And LLMs look, looks not only on, I don't know, search engine, Reddit, mm, and other stuff, but they goes to your case studies, find relevant case studies and say, "Okay, these guys are the best one because they have the same case. These guys are f- are better for, I don't know, for crypto wallets," and so on.

So that's why I would say that we, found our positioning and clear our clients. We started to work on it, uh, in many areas just to build a brand. and I would say that this is the … This approach helped to, to be mentioned in LLMs. David Quaid SEO: You're saying that the LLMs went around the entire internet based on what you said about your brand, they believed that you were the best for crypto marketing, even though they ne- they had no way of testing you against any other agency.

They just - You said that you were the best, and they were like, "They're the best. That's it." Vlad: would say that they rely on all information from internet, including reviews, including case studies, including, including, I don't know, media and so on David Quaid SEO: So because a customer said you were the best, that means you're objectively the best Vlad: Um, you mean that how LLMs decide that we are the best or what? David Quaid SEO: So, so you said that you, you created your brand entity information, right?

Using and publishing information on the internet, and the LLMs were able to come and read that and then just ascertain that you were the best. They just, that's it. Because you had the best reviews or more reviews or whatever, they, based solely on that information that they couldn't validate, they decided you were the best. That's essentially what you're saying?

Vlad: So actually they gathered all the information from the internet, f- I mean, from the search engines, let's, let's say from, from, from the web. then they've, they've d- I think that they've done for the same for other competitors, for the other marketing agency focused on crypto, then they on their weights, on their numbers which, which company fits better David Quaid SEO: But they do this for cars, motorbikes, airplanes, travels, hotels, bed and breakfasts, fast food restaurants, pizza, recipes, agencies, paid search agencies, social media agencies, um, trucking companies, and they do this for the whole world.

They go around and they assess every single brand, and they go, every single brand, this is the best in category. And they do this either at training or along the way, and they go, because of our weightings, whether it's reviews or PR or how many mentions you have or how many reviews you have, they take every single category. Crypto, because I mean, if they're doing crypto, they've got to do PR agencies, right? So there are PR agencies in England, Wales, France, Scotland, Poland, where you are, right?

South Africa, Malawi, Kenya. So for every company, like there's like what? 5 million small companies in America. They've got every single brand, whether you're a plumber, electrician, a landscaper.

They go and look at every single brand, and they've built a database of all of the brands, and they've gone like, "This is the best in every single brand Vlad: They get this information from their memory, from search engines as well. Uh, and additionally, they also can grab some information and to get the, uh, the most important information which is required for the exact request. If you are looking for the best crypto marketing agency for crypto exchange, let's say, so it means that they not just rely on the top 10 in Google, on ranking Google, we- who are there in top 10 of Google.

As I mentioned, they also looked at the case studies, which Google doesn't, and I think that, yeah, they looked at the case study, they look at the reviews, did in the reviews were mentioned, I don't know, the same cases and so on. So they got, I would say, more information and than Google does. So, and based on this information David Quaid SEO: They have more, they have more information than Google has Vlad: I would say that they have the same information, but they, they do analysis of this information better than Google David Quaid SEO: Okay, so they - So if I go to an LLM and I type in who are the top SEO agencies, and I see Google return certain amount of results, are they then getting data on all of the brands that Google excluded?

You're saying that they actually… So, so why use Google if they already know what the answer is? I don't understand it Vlad: They don't know the… So, you know, like, there are different types of, um, uh, different types of approaches in LLMs. So, you know, like, uh, it could be a short answer from their memory. It could be using web search.

So it could be research when they do, mm, deeply, searching for d- for, for additional information. So ba- based on this, you can divide. But if you, let's say, if you use, I don't know, Opus, uh, 4.8, so, and you ask, "Okay, what's the best crypto marketing agency?"

They definitely get this information from the search engines, but the next step will be analyze this information, not just read who are the best, but get some additional information about, as I mentioned before, case studies and so on, and Google do it. So, you know, like, you just see the list of links and that's all David Quaid SEO: So you're saying then that the - depending on how hard it works, the, the best agency changes? Vlad: Uh, it depends on your request and how deeply you want to evaluate.

Yeah, definitely, yes David Quaid SEO: But if you say you want the best agency and then you say, "I want you to do the best agency, but I want you to do more research," it changes. But it already knew who the best agency was because it did the research Vlad: It could be slightly changed, yeah David Quaid SEO: Okay. so best is kind of like subjective Vlad: Uh, yeah, you know, you, you can add some additional information and you could change your, mm, your answer. Yeah David Quaid SEO: So if it, if it's already done this research - 'cause, like, what you're saying is it has to research every single brand, which means it has a database on every single brand, then why is it going to Google?

Like, it doesn't make any sense. Like, so for example, if I say, uh, "Give me the best crypto networks," and I think you, you said you arrived at this conclusion because you were ranking in Google and then you dropped in Google, but you remained. Whereas what I would say is that if you look at the query fan-out, the query fan-out probably included like something like a year or perhaps a different page that quoted you was still left in the query fan-out, and then you would be recommended in the list.

What seems to make more sense is that if you - again, every time we run a search, we see a query fan-out. We never see brands in the results that aren't mentioned in the pages that are in the result set. But I've never seen a brand sort of like appear from nowhere. At the same time, if I can go to Google today and do a search for best SEO agency in Canada, and I've never built a page for s- best SEO agency in Canada, and then I write the page best SEO agencies in Canada, and I put my agency there, and then I run a query, and it goes and does a query fan-out, and Google returns one of my pages with primary position in it, and then I appear in the LLM results, how am I appearing in the LLM results if the LLM is doing all this brand research and it's never mentioned me before?

Vlad: actually, David, I didn't say that LLMs have all brands and all information about the brands. so, David Quaid SEO: Okay Vlad: so they use search engines, so, but they use it differently and evaluate information differently. let's say that, I don't know, you have a SEO agency in New York, right? So, mm, if you, for example, search for SEO agency in New York, I don't know, and it will be top ranked in Google.

And if you say that this informa- and you've worked on, I don't know, Reddit, or you've worked on other stuff like in SEO to, that your website, be ranked in, um, Google. So if I go right now in Perplexity, ChatGPT, Claude and ask the best agency in New York, am I right that I'll see the same result as, uh, I have in Google? David Quaid SEO: No, because one, the, the query and the prompt changes. Two, you have like a, a history with the LLM.

That's not what I'm arguing. What I'm ar- arguing, what I'm saying is that I don't see how the LLMs are building, um, information on certain brands, and then it's only some brands like that match special criteria. But then I can go and do a search for a recommendation for a category, and it doesn't recommend the brand, and then I can create a page that matches certain query fan-outs, and then that's inserted in the LLM all of a sudden that wasn't mentioned by the, by the LLM yesterday.

How is it that the LLM is deciding on the brand if it was missing from Google today and it's visible tomorrow? That's - I'm just asking what the rules are Vlad: So they use, uh, their internal algorithm. I can't say, you know, like exactly how it works, but at the same time they have the memory, so they can use a web browser right now, but the final decision and, uh, the final decision, what to show is on its side David Quaid SEO: But if the, if the search - if, if, if the brand wasn't in the search engine yesterday, and then you up- you use the same prompt you used yesterday, and now the brand is mentioned in the query fan-out, and now the LLM produces that brand, how is it that it's using its own criteria if it's changing because Google has changed?

Vlad: So, you know, like it, it gets information from Google for sure. So I don't say that all from Bing or from, I don't know, Brave, so from different search engines. point is that, uh, it's not the same like in SEO. not the same ranked like, uh, you see in Google.

So it's different David Quaid SEO: Yeah. So that's what I'm saying. If yesterday and - so for this whole year, right? If I take an, if I take best EEAT checker, right?

I was never ranking in best EEAT checker. And I'm not, I'm not a - I don't have an EEAT checker, right? My - I'm not saying that I appear in the list of EEAT checkers, right? My brand was never included in it, and the information on my page was never cited in that page, right?

'Cause it's ac- LLMs aren't actually designed to recommend brands. It's just what they do. They just get information, and if there's brands in a page, it's returned. They, they don't one brand over another.

They don't actually do any research because I can… let's, let's not, let's - EEAT checker is a bit difficult. Let's do EEAT expert, right? So the top EEAT experts, right? 'Cause Edward and I appear in that list sometimes, right?

Because there's a lot of people that are in the EEAT expert list. I was never in the list because I was never an EEAT expert. But I created a list of EEAT experts, and I put in people that weren't EEAT experts. the following day, after being included in, in, in Google, when I went to ChatGPT or Perplexity, and I'm not saying it, it's there for every single prompt because there's a query drift, and I don't have the time to - I don't actually want to be an EEAT expert.

It now recommended Edward, me, and Lars Lofgren, and we were both on the record of saying that EEAT is completely made up, right? So if it's doing all this research and it says, "This is the best brand or the best expert," and we have a 15-minute video of Lars going on a, an F-bomb tangent saying EEAT is rubbish, why then is all of the LLMs recommending them if we're all saying it's rubbish? Vlad: So I would say that if you do a deep research, so they definitely say that, uh, you are newcomer in this market, let's say, and please double-check it David Quaid SEO: Doesn't say I'm a newcomer.

I, it says I have 20 years experience in EEAT as an EEAT expert Vlad: Uh, so I mean, if you are, if you are based just on, on one article or one blog post or something like this, so and you know, like we've made such kind of test before, if I want that, um, um… So you can, you can insert in LLMs any information you want through search engines, that's definitely true, and you could be EE-, uh, you could be any expert you want, but the point in long term doesn't work. Yes, today, uh, we will see that you David Quaid SEO: Does Vlad: an David Quaid SEO: it?

Vlad: but if you don't prove this like day by day, week by week, so you will disappear. do the same for - we did the same test, you know, like before David Quaid SEO: Okay, so have you done a te - So have you done a test like this, right? Um, I find this one really interesting. Top SEOs to follow on Reddit, right?

I would expect if I was asking an LLM who are the top SEOs to follow on Reddit, an agent would go through Reddit, come back and tell me that these people are voted the most, or talked about the most, or speak the most, or get the most upvotes. But when you look at the query fanout, whatever search engine it goes to, it finds the most relevant page, and the most relevant page is actually a Reddit that says these are the top SEOs to follow. So it's not going to Reddit and going like, "Who are the top SEOs?"

It found a pa - or it's given a page that says, " Here are the top SEOs to follow." Now, it also gets 10 other pages. of them aren't even from Reddit because Google rarely returns more than two results per page. And the people that are on the list may not even be on Reddit.

So it might return me, it might return Edward, it might return, um, Lily Ray, it might return Joey Hawkins, who is on Reddit and is amazing SEO as well. and it'll give all these people like Barry Schwartz, and Barry Schwartz isn't active on Reddit. He hasn't been active on Reddit, like, in three years. And it's been doing this for the last eight months.

How is it saying that these are the top SEOs to follow on Reddit one, they're not even active, and two, they're only in one page on Reddit that just matches the search phrase? Like, where, where are they doing the research? Vlad: actually, uh, mm, we definitely need to make a fact-checking if you get information from LLMs. So, and by research, I mean that they, uh, g-go to search engines and look not just on Reddit or blog post maybe they found, but to try to find more.

I mean, maybe reviews, maybe other mentions David Quaid SEO: But they literally, Vlad: gather all the David Quaid SEO: they literally just take th- they take that one page. You can literally visit the page and you can see it in the list, and they literally go in the same order that are on the page that one person wrote Vlad: Okay. But, uh, you know, like, m- maybe it happens today, so I need to, to see this page David Quaid SEO: For nine months Vlad: Yeah. but, uh, mm, you know, like I, I would say is that if you're talking about long term, and let's say I'm not sure that it will be the same for, for next year, for next six months, for next three months.

As a brand, David Quaid SEO: Next year, another page, another page could change, right? I'm saying that it's recommending people that aren't even on Reddit. It's only recommending them because they're on that page, and before that page ranked, it never mentioned them. That's what I'm trying to say.

Like, I can a list and put it into it, and depending on that query fanout and the drift in the query fanout and the pages that are - that come up in each term, it's pivoting all the time. So if it's doing all of this research, why is it showing people that don't exist on Reddit as the best people to follow on Reddit? And it has been doing it consistently because I tweet about it once a week on, on X Vlad: So, you know, David, I think that it's a good example that you can cheat LLM and insert any information you want.

So even people who are not David Quaid SEO: But if I can cheat an LLM, but you said it's doing all this research and it's understanding all this brand equity, so how is it doing that and not doing it at the same time? I don't understand that Vlad: let's say GEO could be a cheating LLMs and to put the information you want. So that's why this is a good example that, you know, like uh, GEO works. I mean that if you want to insert any information in LLMs, it's doable even if you are not expert in Reddit David Quaid SEO: Okay, but I'm, I'm saying this isn't a question of like geo versus SEO.

I'm, I'm just trying to understand if you're saying that it does all this research. I'm saying, but if I change the constituent pages and the pages are changed because Google has changed it, it sounds like to me that Google is picking the results from which the LLM is making its deductions. Edward Sturm (2): Well, you're making a - I wanna say you're making an interesting point. You might - I'll give an example of a press release.

So it's, it's well known that you can buy a cheap press release and that will - and you, you put a keyword in the title of the press release, and then that will actually, the LLM will see the press release, it'll synthesize what's in the press release and kind of quote it verbatim or say whatever the press release says. And to an, to a user who says, who sees the press release, it actually might be like you might see it syndicated on Yahoo Finance or something. And to the user it, it'll say at the top, you know, "This is a press release and, and this is like a paid press release."

And, and so the user might be like, "Oh, this is like a press release and I shouldn't really trust this." But the LLM might not even say that. The mi- the LLM might just say what's in the press release. And, and, and so that's kinda like one of the differences that you're getting at.

Is that, is that correct? Vlad: So actually I would say that, yeah, LLMs can get the information from press releases, but at the same time users don't read it. Yeah, that's true. Edward Sturm (2): Like, well, it's at, it, it's at the user, it's at… Yeah, if the user wa- if the user saw the press release, the user would know that it was a, a paid press release that was syndicated to another website because it would say it at the top.

But ChatGPT is just gonna say what's in the pre- Sometimes ChatGPT might say what's in the press release without saying that this is actually a paid press release and that you shouldn't trust it Vlad: Mm, that's true. So sometimes it happens Edward Sturm (2): And, and that is something that, that like SEOs know how to do. It's like, oh, this is… We can, we can put out a pr- Like, uh, this is actually an older… This was like, uh, Harpreet, who's been on the show many times before.

He, when he shared this, he said, uh, Harpreet Singh, he, he said this was an older SEO tactic that SEOs used to use all the time that stopped working as well just because, like, like we- people know that this is a press release. But now that ChatGPT or AI Overviews are quoting the press release without even saying that it's a press release, it's actually sometimes it's, it's more worth doing David, you were g- you were saying something? David Quaid SEO: Yeah, and I - Yeah, I - So I, I - This is a great point because this is exactly where I'm getting to, right?

If I - Let's say I see, um, my competitor putting up a press release saying that they're the better SEO agency or the better EEAT expert, I can take my SEO skills and push their press release out of the way, and then I'm mentioned by the LLM. And what I think is fascinating is if you look at the number of press releases, re- press releases, and not - let's not even talk about syndication, right? 'Cause like one press release can be syndicated 1,800 times, right? If I can push that press release out of the way in Google and I do the same prompt in, in the LLM, and then because the, my competitor's press release is no longer quoted and I am, it doesn't seem to me that the LLM is picking the press release.

It sounds to me that Google is, or the search engine, is giving it the press release and it changes its opinion based on what it's fed. That doesn't sound to me like a system that's researching it Vlad: Mm-hmm. Um, say that, you know, like, um, you, if you compare the two different - the same request in Google and in LLMs, and you ask a person, "Okay, which answer you like more?" I'm not sure that, that, uh, people, users just say, "Okay, answer from Google is better than from ChatGPT."

What does it mean? Uh, it mean that, okay, the final answer from ChatGPT or, I don't know, other LLMs, so they like it more. It could be, I don't know, sponsored, it could be, I don't know, uh, faked and so on. But at the end of the day, so it's the, a person who uses, uh, LLMs, so the… I would say that they like it more Edward Sturm (2): Gentlemen, I wanna, I'm gonna wrap with, with one question.

We've been… This has been very good. We've been on for two hours. I'm gonna wrap with this, with the, with a question, and, and, and I'd like an answer from each of you. David, uh, and, and the q- this, the question is for, for both, but we'll get a cool answer from each.

Um, David, a, a person comes to you and they say, "I want to, I want my brand to be recommended by ChatGPT. Give me the top two things to do." What do you tell them? David Quaid SEO: So I would say take a look at what the most common questions asked are.

Uh, see where you're ranking and not ranking. Uh, go to the, the, the engines, the LLMs, put in the prompt, and then look at the query fanout. So Vlad was saying, for example, earlier that his search - he was no longer ranking Google, but he was found in ChatGPT, and that's because the query is different, right? Remember I had that cartoon with like the prompt is not the query, right?

So basically, you could be - you could do - ask for what is the top crypto site or best crypto site, but the LLM is saying, "Who are the best recommended networks for twenty twenty-five?" For example. And even though you don't rank in Google for the first phrase, you actually rank in Google for the second phrase, you will be recommended in the final result set. So all we have to do is find the difference between the prompt and the query, and once you rank for the query, you'll be regurgitated by the LLM about ninety percent of the time.

Edward Sturm (2): Yivara Vlad: So from my side, I would say that the first point we need to make an AI visibility audit based on the hundreds of different requests to understand where we are and where our competitors are. So then, uh, to clearly understand like our position on brand entity, to see the gaps between competitors if they are higher in LLMs than our brand. So, and then just to work on it on, uh, these gaps on its brand entity. Yeah.

And do it like, uh, week by week Edward Sturm (2): Gentlemen, thank you so much. Vlad, this is s- since this is your first time on the show, where should everybody find you? Vlad: Uh, so actually I would say in, in LinkedIn Edward Sturm (2): Yep, and all of your, all of your links will also be in the description. So on LinkedIn and you have Stive.

ai. David, you've been on many times. Everybody knows you. Uh David Quaid SEO: Yep.

Everyone knows if you look for me, you can find Dennis Quaid and all his movies, and it's 'cause, 'cause Google's so smart. Vlad: Yeah. Edward Sturm (2): This is episode 1,087 of the podcast. 1,087 days in a row doing this show.

If you watch this on YouTube, thank you so much for watching. If you listened on Spotify or Apple Podcasts, thank you so much for listening, and thank you again to my wonderful guests, and I will talk to you again tomorrow. Bye now.

Related episodes across the Index

Other episodes covering the same guests and topics, from across The B2B Podcast Index.

  • Episode 029 AI and the Rise of the Superpowered SoloAI Tools for Practicing Lawyers · on Claude88 / 100
  • Wins Above Replacement: The New Way to Judge FoundersVenture Unlocked · on ChatGPT88 / 100
  • Paul Graham On Startups, Ambition, and Great FoundersY Combinator Startup Podcast · on ChatGPT88 / 100
  • The AI-Native Law Firm, with Ryan Walker of General LegalMeeting of the Minds · on Claude88 / 100
  • Adamma Ihemeson on Building Autonomous AI Systems You Can TrustBack on T-R-A-C-K · on Claude84 / 100
  • Athos Commerce CMO Gary Lombardo on why only 14% of shoppers see one consistent brandThe Agile Brand with Greg Kihlström® · on Generative Engine Optimization (GEO)80 / 100

More from The Edward Show

All episodes →
  • An iGaming SEO Reveals How He Manipulates ChatGPT Recommendations81 / 100
  • Google Can't Stop Fake DMCA Attacks - And SEOs Are Getting Wiped Out71 / 100
  • The Digital PR Playbook That Earns High-Authority Links (Most SEOs Never Learn)80 / 100
  • If I Started Over at 22, I'd Use This Playbook to Pay NYC Rent in 90 Days66 / 100
  • How the GEO Playbook Is Challenging SEO With Paid Citations, Reddit & AI Search75 / 100
Explore the best B2B Marketing podcasts →
All The Edward Show episodes →