The GEO Myth: Why AI Search Is Just SEO in Disguise
Chapters9
A creator explains six months of testing various SEO tactics for geo optimization without a clear signal of what influences rankings, noting that third-party mentions seem more impactful but measurement remains murky.
AI search is just SEO in disguise, driven by how LLMs crawl the web, perform query fan-out, and rely on grounding signals rather than training data alone.
Summary
Edward Sturm dives into the mystery of what actually moves the needle in geo-focused AI search. He flags a thread where practitioners report inconsistent results after six months of testing FAQ sections, citations, and reviews, with third-party mentions often standing out. The discussion centers on the concept of a query fan-out—the searches an LLM performs to gather signals before answering—and grounding queries that validate sources. Web linker, a notable SEO subreddit moderator, demonstrates how an LLM can fetch fresh content and rank in minutes by leveraging the query fan-out rather than stats from its training data. The episode also touches on real-world implications: brands, geo agencies, and even startups are chasing visibility through AI search, and tools like Claude, Claude’s web-search behavior, and prompt-research workflows (via SEO Pub’s and PromptWatch integrations) illustrate the mechanics behind AI-cited content. Sturm emphasizes that AI search benefits from strong SEO foundations—especially reputation management, topical authority, and landing/product pages that actually get cited by chat models (reported as 15.2% and 14% respectively by PromptWatch co-founders). He also points listeners to practical strategies like compact keywords and landing-page SEO as the fastest path to citation-driven rankings in 2026. The segment closes with a nudge that the geo industry is being reshaped by how search engines see and cite brands, not just by on-page optimization alone.
Key Takeaways
- The query fan-out is the initial set of web signals an LLM uses to build an answer, e.g., the top geo agencies NYC and related brand terms.
- Grounding queries validate sources during AI construction, and brand names can become part of the grounding process (e.g., Mimvi, Primary Position).
- Claude and other LLMs show web-search traces and grounding steps, debunking the myth that AI relies solely on training data for recommendations.
- AI search visibility platforms like PromptWatch report that landing pages (15.2%) and product pages (14%) are frequently cited by chat models.
- Compact Keywords and SEO landing pages are shown as effective strategies to increase AI-cited content and direct brand relevance in 2026.
- There is strong industry chatter about a “geo lobby” and disinformation about AI training, but practical demonstrations reveal that AI searches mirror conventional SEO tactics plus citation-building.
- Episode 9997 reinforces the habit of daily podcasting by Edward Sturm, underscoring consistency as a competitive edge for content-based SEO.
Who Is This For?
Essential viewing for geo SEO practitioners, AI marketers, and brand managers who want to understand how AI search citations actually happen—and how to structure pages that get cited by chat models.
Notable Quotes
""The AI, the LLM will go to a search engine and will do a search and gather the data there and then put together an answer.""
—Web linker explains the basic mechanism of AI search behavior.
""Grounding queries are used to validate results to compare sources to enrich the answer.""
—Describes how grounding signals refine AI outputs.
""I think I could get a blog post to rank by this evening for this phrase and put in whomever I want.""
—Illustrates the perceived manipulability of AI-generated rankings.
""Landing pages are getting cited by chatbt 15.2% of the time. Product pages 14% of the time.""
—Cites PromptWatch data to support a concrete optimization path.
Questions This Video Answers
- How does a query fan-out influence AI-generated search results and rankings?
- What are grounding queries in AI search and how do they affect citation quality?
- Which SEO pages are most cited by ChatGPT and similar models when answering queries?
- What role do Claude and other LLMs play in real-time web searches for SEO?
- How can geo agencies optimize for AI search visibility in 2026?
AI searchQuery fan-outGrounding queriesPerplexityClaudePromptWatchGeo SEOLanding pagesProduct pagesTopical authority
Full Transcript
six months trying to get cited in AI and I still can't figure out what's actually moving the needle. This was posted to the SEO subreddit. I got tagged in it. Passage alarmed 549 said, "Edward Sturm, please make an episode from this thread." Here we go. So, the original post is this person has been doing generative engine optimization geo work for 6 months. Still can't figure out what is moving the needle. 6 months, three clients testing everything. FAQ sections, content comprehensiveness, citations, reviews. Results are inconsistent, and I can't isolate what is working. The one pattern I keep seeing is third-party mentions mattering more than anything on site.
They do matter, but I can't tell if that signal or noise without reliable measurement. What are people actually finding works, and how are you measuring it with any consistency? All right. So, the top comment is from the one and only web linker, the top moderator on the SEO subreddit who said, "Here is an example. I only created this a few weeks ago. Ergo, it is not from Perplexity's quote unquote training set. I can run a prompt, not be visible, find the query fan out." So, that means the things that the AI is actually searching for to get its answer.
the AI, the LLM will go to a search engine and will do a search and gather the data there and then put together an answer. And so web linker says find the query fan out, find the searches that are being done, write a new blog post and rank in Google in 10 minutes and appear in the query fan out. Then it is impossible for it to be training. And yes, not only am I quoted in the results set, I am in the additional grounding. So this is the screenshot that is shared. The prompt is top geo agency NYC.
The initial query fan out. We see it. Perplexity is searching top geo agencies NYC. Best geographic optimization agencies New York City. Top geo SEO agencies New York geo agency NYC reviews. And then the grounding queries. These are used to validate results to compare sources to enrich the answer. So the grounding queries are top geo agencies New York City 2026, best generative engine optimization agencies NYC, MIMV SEO GON NYC, primary position geo agency NYC, local mighty geo NYC, I pull rank geo services. So it is actually the the grounding queries are brand names now because the initial query fan out got some brands and then is doing research into the brands.
So it's searching things like Mimvi primary position local mighty I pull rank again these are all brands and web linker is saying I'm in the additional grounding queries and so here is the thread that this person asked me to to cover so the followup to web linker is pig the kid who said this is super interesting and not something I have tried before how are you finding the query fan out and can you do it on claude or any LLM you work with web linker said, "I know. I feel I've been banging this drum daily on podcasts X and here for 2 years and I've barely reached 1,000 people." The geo the generative engine disinformation about training is so strong on perplexity.
You click the steps tab. Claude, it's the same thing. For chat GPT, you can use a query fanout tool and web linker links to this one from SEO Pub, which I'll share in the description for that. actually just messaged uh SEO pub because his one had stopped working and so he fixed it. I let him know and then he fixed it and he let me know. So he has this updated search query and reasoning extractor bookmark lits chatpt plus cla and it shows what chatpt is searching with. So web linker continues this is a specifically good example.
I asked claude top SEOs to recommend on the SEO subreddit. Claude openly shows you it is doing a web search. None of the pages returned by Brave Search are actually from RS SEO. This has nothing to do with training. Claude cannot possibly hold every LinkedIn article, blog post, and Reddit post in its quote unquote memory. A lot of people think that if you are getting recommended something by an LLM, it is using data that is already in its training data. And that's not the case. The LLM is doing a web search like you might do a web search and then it is doing grounding queries and going deeper.
Web linker says training is mostly heruristics patterns that it learns mostly sentences that are made up. When you see this you begin to see the smoke and mirror story. I think I could get a blog post to rank by this evening for this phrase and put in whomever I want. Weblinker shares screenshot top SEOs recommended to follow on the SEO subreddit. And you see that Claude is searching the web. So a different user follows up and says, "I think I caught you weblinker on a podcast about this a while back. Felt very stupid afterwards and now I cannot see why this isn't the default narrative.
Is there actually some kind of lobby covering up the query fan out and building a geo industry based on other random things?" And weblinker said this is the default mindset of about 99% of people. You should feel like you're in the top 1% who can actually see this. Listeners of the podcast, we are in this top 1% who know the truth. The truth will set you free. You know it now. You know how to get into LLMs. Weblinker said, "There is 100% a lobby. I was offered money at billable rates plus equity in startups to post for 2 years that SEO was dead and LLMs are independent search engines.
I posted that here and Edward Sturm, that's me, covered it too. I did cover it. That was episode 811, the fake geo movement. How SEO disinfo campaigns target creators. Weblinker continues, "If you look at some of the geo subreddits, they are agencyowned, not even hiding it, and it is all bots that are publishing. It is easily the biggest smoke and mirror trick in marketing." Thanks so much for sharing this. I really, really appreciated hearing this person followed up and said, "I think this is absolutely crazy and deserves about 10x the exposure it is getting. Surely at some point in the next 12 months, brands will look the thousands they have spent on Geo." It's a lot more than that and wonder where the business growth is.
It's like the funding for these geo companies is crazy and the investors don't realize it is SEO that the LLMs are doing web searches. It is really not very different from knowing how to do SEO and knowing how to do reputation management with SEO and reputation management has always been part of search engine optimization. So this person says, "By the way, you're doing a great job of what you could call the quote unquote SEO realism. Keep doing it. Much appreciated." And that's where Passage Alarmed 549 said, "Edward Sturm. Please make an episode from this thread." And you got it.
Last thing that I want to share, this is pretty cool. This is from one of the co-founders of PromptWatch. Promptwatch shows you how you're getting cited in AI. Funny enough, PromptWatch's page title is the number one AI search visibility and geo platform. But, you know, I had uh I had another one of the founders of PromptWatch, Clauss, on the podcast and he's a really nice guy and they are the number one AI search visibility platform. That was episode 856. How AI search optimization is changing Google chat GPT and the web. I would say actually that AI search is changing the web because it's making SEOs focus more on SEO reputation management which they should have been focusing on all along.
It is crazy important. It's also important for doing entity SEO and for building topical authority. When you when your brand is appearing in lots of different places with language about how you are relevant in your brand, it enforces this and it makes it easier for you to rank for keywords within your niche. And when you're ranking for these keywords within your niche, you are getting cited. But what content is getting cited? And that's what I want to share. And this is from another co-founder who said, "Seeing a lot of talk about listicles losing their relevancy in chat GPT citations, but we are seeing the exact opposite.
Over the last 30 days, we've collected data on 1 million citations in Prompt, and the results are clear. Listicles, landing pages, and product pages are the top sources that Chat GBT sites in responses. Geo moves fast, relying on recent and large data sets are what will help you win in 2026." Lily Ray posted this amazing article about how self-promotionalisticles are not doing as well. I covered that as well, which that is a mustwatch episode. Episode 946 of the show, Google is firing shots, self-promotional listicles are getting wiped out. And Lily Ray actually commented in this and said, "Did you check if they were self-promotionalistic?" The co-founder of PromptWatch said, "We did.
There's more data coming soon. We'll see what that says." But I want to share this because landing pages are getting cited by chatbt 15.2% of the time. Product pages are getting cited by chatbt 14% of the time. And if you have been watching this podcast for a while or if you have gotten my course compact keywords, my course shows you how to make SEO landing pages and SEO product pages that are getting cited so well in ChachiPT. This is the best method of search engine optimization. I have been screaming about it for such a long time because SEO landing pages they sell.
Top offunnel articles they don't sell. Top offunnel articles are just getting awareness. And that's what a lot of people think that search engine optimization is. A lot of people think search engine optimization is putting up how, what, why questions so that they can get their so that they can get people to see something on their site and then slowly move them into their funnel. What a lot of these marketers don't realize is people are literally searching for exactly what your brand sells, but they don't know that your brand exists. They are searching for what your brand sells.
And so you put up these SEO landing pages, these SEO product pages targeting these people saying, "Hey, we are the solution for the problem that you are searching for right now. This is why we're the solution. This is why we are good at this solution." And these pages that I that I show how to make in compact keywords, my method is getting cited 15.2% of the time. And then another 14% of the time, total 29% of the time. Come on, that's amazing. I shared this testimonial. Been getting amazing testimonials. I share this one yesterday from Angel who said, "Just a one-man shop getting ahead of GQ, Esquire, Men's Health, and several milliondoll brands.
I hope you will check out the course that is at compactkeywords.com. I keep adding amazing testimonials to it. I just added a new testimonial, fresh one to it last night. It is the best way of doing SEO because it sells your brand. It is also less work because there's less people targeting keywords at the bottom of the funnel. The keywords are less competitive. They require less words per page because searchers are already in heat and getting cited in chatpt 29% of the time. That's at compactkeywords.com. And that is everything that I got for you on this episode of the show.
This is episode 9997 of the Edward Show. 997 days in a row doing this podcast. No days missed. If you watch us on YouTube, thank you so much for watching. If you listened on Spotify or Apple Podcasts, thank you so much for listening and I will talk to you again tomorrow.
More from Edward Sturm
Get daily recaps from
Edward Sturm
AI-powered summaries delivered to your inbox. Save hours every week while staying fully informed.









