← back to all posts

Mueller: "AI Systems Build on SEO." Why the Technical Base Decides GEO and AEO Too

Mueller: "AI Systems Build on SEO." Why the Technical Base Decides GEO and AEO Too

On November 2, Search Engine Journal published Roger Montti's piece on John Mueller's interview with David Iwanow. The line that matters: "a lot of the AI systems, they build on SEO." That's the subject of this blog, so here's what Mueller actually said, and what retellings have been adding on his behalf.

What was said verbatim

On technical SEO. Asked whether AEO and GEO are dead, Mueller answered:

"I think you'll see that a lot of the AI systems, they build on SEO. So from that point of view, I think it's going to continue to be relevant. I think especially technical SEO is going to remain very relevant. It'll be a little bit different because my understanding is AI systems crawl slightly differently. They have different intents with regards to crawling because they want to do it for training versus just like getting the freshest content. But that's just is like another flavor of technical SEO."

Note the second half: AI crawlers behave differently and come with a different intent — some crawling is for training, some for fresh content. But in Mueller's framing that isn't a separate discipline, it's "another flavor" of technical SEO.

On listicles. Asked about "top 10 companies" roundups published to influence chatbots:

"Something comes out and people realize, oh, this actually works particularly well. And then everyone goes off and does it and half the people go off and do it in a particularly bad way. And then at some point the systems catch up and they're like, oh, anything that has a listicle is a sign that it's spam. And it's like the circle closes and you find something else."

He stops short of a ruling: "I wouldn't necessarily say anything that is a listicle is spam... It needs to add value." Then comes the warning worth pinning above the desk:

"So if you go all out and say, my whole strategy is listicles, then at some point that's going to switch off and you're left with nothing really. Building up a serious, stable foundation for a website, I think that's the most critical part."

On vendors. On questionable techniques sold as AI work:

"It's like you're paying consultants and they do, I don't know, keyword stuffing. Or back in the early days, it's like white text on white background. And it's like, it works for a bit, but at some point, it doesn't work. So I think as a small business, you have to always question what your vendors are doing."

What Mueller did not say

Retellings of this interview have been circulating for a week, and they've picked up a lot that isn't in the source. In the SEJ piece Mueller says nothing about RAG pipelines, nothing about buying junk links, nothing about "agencies selling AI SEO," and nothing about rendering or indexing. Listicles are subtler: the self-promotion framing comes from the article's author, who defines listicles as "top companies" lists where the company compiling the list puts itself first. Mueller's own answer deals with listicles in general and never touches the self-ranking part.

Meanwhile the interview contains a detail the retellings dropped: just before this, Mueller said there's no reason to build separate markdown versions of pages for Google's AI Search. For anyone currently debating llms.txt and machine-readable content duplicates, that's a far more practical piece of news than another round of "lists are spam."

And the conclusion about old tactics coming back — cloaking, directories, reciprocal links — belongs to the article's author, Roger Montti, not to Mueller. Montti offers it in his own name, on the strength of 25 years in the field.

Why I agree

My own data has confirmed this twice, once crudely and once precisely.

The crude one. At a federal media site, Google couldn't crawl part of the pages for 89 days: Search Console reported up to 78% failed crawl requests. The contractor blamed state-level blocking. There were three causes, all on the server: an IPv6 address that existed only in DNS, nginx serving it from the wrong block, and Imunify360 cutting inbound traffic from the US. Access came back an hour after the geo-block was lifted. Full write-up on Habr (in Russian). No amount of "AI citability" work would have mattered there — the crawler never reached the content.

The precise one. On the anastasiya.studio store I matched the new Search Console Generative AI export against the regular performance report for the same period. Of the 419 pages that received impressions in AI Overviews and AI Mode, 74% rank in the top 10 of classic search, and those pages account for 92.6% of all AI impressions. Not a single page appeared in AI answers while being absent from organic results. That's exactly what "build on SEO" looks like in one store's numbers.

The reverse holds too: ChatGPT goes off to cite a competitor when prices and specs are hidden behind JavaScript — covered with data from two studies.

What to do about it

If AI systems build on SEO, the list of work isn't new, but the priorities inside it are different.

One caveat, so I'm not read too broadly. "Technical SEO stays relevant" does not mean "GEO and AEO are unnecessary." Brand mentions on external platforms, content structured so it can be quoted, linked author profiles — none of that is on a technical checklist, and all of it affects citation. The technical base doesn't replace that work; it's the condition under which that work pays off.

The Search Engine Journal piece is worth reading in full: it quotes Mueller at length, so you can see where he stops and the author begins. The page also embeds the interview itself, if you would rather hear the wording in full.

Maxim Safianov
Maxim Safianov

I work at the intersection of technical SEO and software engineering: helping sites stay visible for classic search and AI answers alike, and building the Python tooling that automates it.

No comments yet — be the first to share your thoughts.

SEO

Did Your Rank Tracker Show a Collapse in Late September? Google Likely Served It a Fake SERP

Nozzle analyzed 27.5 million SERPs and found that from September 20 Google began serving rank trackers substituted results — no errors, no captcha, outwardly normal. At the September 29 peak, 27% of SERPs were poisoned, and the tracked site fell out of the top 10 in 58.6% of them versus 2.7% in the baseline. Here is the data, a three-minute test to tell a poisoned SERP from a real drop, and why I stopped tracking positions.

14 Oct 20266 min
Python

Chris Green Built a Chrome Extension for This. In Orakul It's Just Text

On September 30, Chris Green wrote in Search Engine Journal that not every SEO task needs a frontier model: exact operations belong in code, light interpretation in a small local model, and the expensive frontier model should be called only for judgment. I built a product on exactly that principle. Here's how the same architecture gets assembled in text, with no code, and where my stack falls short of his design.

01 Oct 20266 min
GEO

Google's Top 10 Still Feeds AI Answers. Three Signals the Rules Are About to Change

The common view is that AI answers have already changed search. Under the hood it is still the same two-stage ranking pipeline. I checked this on data from anastasiya.studio: 74% of pages with impressions in AI Overviews and AI Mode rank in the top 10 and collect 92.6% of AI impressions. But Google has already published the theory that removes that pipeline, started paying for contributions to AI answers, and opened a Search Console report for them. Here are the three signals and what to do right now.

17 Sep 202610 min