Google\u2019s Danny Sullivan and John Mueller\u2019s Search Off The Record podcast offered guidance to SEOs and publishers who have questions about ranking in LLM-based search and chat, debunking the commonly repeated advice to \u201cchunk your content.\u201d But that\u2019s really not the conversation Googlers should be having right now.<\/p>
SEO And The Next Generation Of Search<\/h2>
Google used to rank content based on keyword matching and PageRank was a way to extend that paradigm using the anchor text of links. The introduction of the Knowledge Graph in 2012 was described as a step toward ranking answers based on things (entities) in the real world. Google called this a shift from strings to things.<\/p>
What\u2019s happening today is what Google in 2012 called \u201cthe next generation of search, which taps into the collective intelligence of the web and understands the world a bit more like people do.\u201d<\/p>
So, when people say that nothing has changed with SEO, it\u2019s true to the extent that the underlying infrastructure is still Google Search. What has changed is that the answers are in a long-form format that answers three or more additional questions beyond the user\u2019s initial query.<\/p>
The answer to the question of what\u2019s different about SEO for AI is that the paradigm of optimizing for one keyword for one search result is shattered, splintered by the query fan-out.<\/p>
Google\u2019s Danny Sullivan and John Mueller took a crack at offering guidance on what SEOs should be focusing on. Do they hit the mark?<\/p>
How To Write For Longform Answers<\/h2>
Given that Google is surfacing multi-paragraph long answers, does it make sense to create content that\u2019s organized into bite-sized chunks? How does that affect how humans read content, will they like it or leave it?<\/p>
Many SEOs are recommending that publishers break up the page up into \u201cchunks\u201d based on the intuition that AI understands content in chunks, dividing up the page into sections. But that\u2019s an arbitrary approach that ignores the fact that a properly structured web page is already broken into chunks through the use of headings, HTML elements like ordered and unordered lists. A properly marked up and formatted web page should already be formatted into logical structure that a human and a machine can easily understand. Duh\u2026 right?<\/p>
It\u2019s not surprising that Google\u2019s Danny Sullivan warns SEOs and publishers to not break their content up into chunks.<\/p>
Danny said:<\/em><\/p> \u201cTo go to one of the things, you know, I talked about the specific things people like, \u201cWhat is the thing I need to improve.\u201d One of the things I keep seeing over and over in some of the advice and guidance and people are trying to figure out what do we do with the LLMs or whatever, is that turn your content into bite-sized chunks, because LLMs like things that are really bite size, right?<\/p> So we don\u2019t want you to do that. I was talking to some engineers about that. We don\u2019t want you to do that. We really don\u2019t. We don\u2019t want people to have to be crafting anything for Search specifically. That\u2019s never been where we\u2019ve been at and we still continue to be that way. We really don\u2019t want you to think you need to be doing that or produce two versions of your content, one for the LLM and one for the net.\u201d<\/p> <\/blockquote> Danny talked about chunking with some Google engineers and his takeaway from that conversation is to recommend against chunking. The second takeaway is that their systems are set up to access content the way human readers access it and for that reason he says to craft the content for humans.<\/p> But again, he avoids talking about what I think is the more important facet of AI search, query fan-out and the impact to referrals. Query fan-out impacts referrals because Google is ranking a handful of pages for multiple queries for every one query that a user makes. But compounds this situation, as you will see further on, is that the sites Google is ranking do not measure up.<\/p> Danny Sullivan next discusses the downside of optimizing for a machine, explaining that systems eventually improve that usually means that optimization for machines stop working.<\/p> He explained:<\/em><\/p> \u201cAnd then the systems improve, probably the way the systems always try to improve, to reward content written for humans. All that stuff that you did to please this LLM system that may or may not have worked, may not carry through for the long term.<\/p> \u2026Again, you have to make your own decisions. But I think that what you tend to see is, over time, these very little specific things are not the things that carry you through, but you know, you make your own decisions. But I think also that many people who have been in the SEO space for a very long time will see this, will recognize that, you know, focusing on these foundational goals, that\u2019s what carries you through.\u201d<\/p> <\/blockquote> I have known Danny Sullivan for a long time and have a ton of respect for him, I know that he has publishers in mind and that he truly wants for them to succeed. What I wished he would talk about is the declining traffic opportunities for subject-matter experts and the seemingly arbitrary garbage search results that Google consistently surfaces.<\/p> Google is intentionally hiding expert publications in the search results, hidden away in the More tab. In order to find expert content, a user has to click the More tab and then click the News tab.<\/p> This search was not cherry-picked to show poor results. This is literally the one search I did asking a legit question about styling a sweatshirt.<\/p> Google\u2019s AI Mode cites the following pages:<\/strong><\/p> 1. An abandoned Medium Blog from 2018, that only ever had two blog posts, both of which have broken images. That\u2019s not authoritative.<\/p> 2. An article published on LinkedIn, a business social networking website. Again, that\u2019s not authoritative nor trustworthy. Who goes to LinkedIn for expert style advice?<\/p> 3. An article about sweatshirts published on a sneaker retailer\u2019s website. Not expert, not authoritative. Who goes to a sneaker retailer to read articles about sweatshirts?<\/p> Had Google defaulted to actual expert sites they may have linked to an article from GQ or the New York Times, both reputable websites. Instead, Google hides the high quality web pages under the More tab.<\/p> This whole thing about GEO or AEO and whether it\u2019s all SEO doesn\u2019t really matter. It\u2019s all a bunch of hand waving and bluster. What matters is that Google is no longer ranking high quality sites and high quality sites are withering from a lack of traffic.<\/p> I see these low quality SERPs all day long and it\u2019s depressing because there is no joy of discovery in Google Search anymore. When was the last time you discovered a really cool site that you wanted to tell someone about?<\/p> Garbage on garbage, on garbage, on top of more garbage.\u00a0Google needs a reset.<\/p> How about Google brings back the original search and we can have all the hand-wavy Gemini stuff under the More tab somewhere?<\/p> Listen to the podcast here:<\/strong><\/p>
Avoids Talking About Search Referrals<\/h2>
Focus On The Big Picture<\/h2>
Let\u2019s Talk About Garbage AI Search Results<\/h2>
Subject Matter Expertise Is Missing<\/h3>
How Google Hides Expert Web Pages<\/h4>
![\"How](\"https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/01\/google-buries-legit-sites-970.png\" "\\"Google")
<\/p> Google\u2019s AI Mode Promotes Garbage And Sites Lacking Expertise<\/h2>
Screenshot Of Google\u2019s Garbage AI Results<\/h4>
![\"Google](\"https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/01\/garbage-search-results-327.png\" "\\"Google")
<\/p> Google Hides The Good Stuff In More > News Tab<\/h3>
Screenshot Of\u00a0 Hidden High Quality Search Results<\/h3>
![\"Google](\"https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/01\/expert-articles-415.png\" "\\"Google")
<\/p> GEO Or SEO \u2013 It Doesn\u2019t Matter<\/h2>