preloader image
May 18, 2026
AI, SEO

Busting the Mythbusting

GEO is just good SEO

We’ve said it before, and we’ll say it again, SEO is not dead. GEO (or AEO, or AI SEO) is just SEO best practices. Anyone who claims that their SEO isn’t enough for GEO isn’t paying attention, or possibly is just doing SEO wrong (they’re possibly still trying 10 year old black hat SEO techniques). This message is the crux of what Google have (finally) officially announced, and, as far as we’re concerned, Google are telling the truth about what it takes to feature in their own Generative AI (gAI) search features (notably AI Overviews and AI Mode).

However, it is important to recognise the fact that Google are not the only game in town, and that their advice differs from existing advice from other gAI search tools. Bing being the most obvious, but also Chat-GPT, Claude, etc.

“Foundational” SEO

The most important part of Google’s new advice is reiterating their age-old advice to place the user first. Don’t create content for the search engine, create content for the end user. There’s additional clarification about producing “non-commodity” content, i.e. specific, authentic, and doesn’t essentially repeat the same old information found on dozens of other websites. Content should provide a unique point of view and, where possible, provide specifics—real-world examples trump vagueness.

Google also repeat their advice to not rely on AI to write your content without human input. Too many sites have faced “Mount AI”, posting tens, hundreds, or even thousands of pages of content all generated by AI, only to see their rapid growth in search visibility drop off a cliff when Google’s system catches on to the fact that said content is all junk.

Myth Busted?

Perhaps the most controversial (at least amongst SEO professionals) part of Google’s new article is the section on mythbusting generative AI search. They listed five things that you don’t have to do in order to be found by Google’s gAI search tools. These need to be addressed one by one.

LLMS.txt files and other "special" markup

Google say that you don’t need an llms.txt file (a text file added to your website to signpost important pages for AI crawlers). There may be some merit to this, as large-scale studies have shown that these files are roundly ignored by everyone. However, this is a developing technology, and it’s not going to hurt to just create and upload an llms.txt file.

Interestingly, Google themselves have added an llms.txt file to their developer documentation, and Shopify recently added one to every shop on their platform (which includes a cheeky advert for their own services). The same goes for markup or markdown pages specifically made for AI. While these might not have any effect, it’s not going to hurt to have them, especially if your website is using JavaScript to render the front end of the site for the end user, as this can be tricky for some AI bots to crawl.

“Chunking” content

Anyone selling GEO services has gone on and on about “chunking” content for digestion by AI. Basically, splitting your text up into easily manageable sections. Google have said that this isn’t necessary, and that their crawlers can handle text just as well as a human, which is who you should be making pages for.

Honestly, this is accurate, but the crux of the matter is that you should follow age-old editorial guidelines for readability. The human eye is put off by walls of text, so your articles should already be “chunked” by the simple expedient of using paragraphs.

Rewriting content just for AI systems

This myth is Google simply repeating themselves with the “write for humans, not search engines” advice that has been their mantra since their founding. Yes, you don’t have to rewrite your content to target every fan out query and long tail keyword. That doesn’t mean that you shouldn’t review your content and maybe rewrite it, at least a little.

Several tests have shown that AI has a tendency to take things literally. Some horror stories of AI telling users to do silly things like eating rocks or jumping off the Golden Gate Bridge have circulated. And while AI has (hopefully) developed past these sorts of mistakes, you should ensure that there’s no ambiguity in your content. It’s getting better at identifying sarcasm and irony, but it’s best to make sure that it is very obvious.

Seeking inauthentic “mentions”

Mentions were introduced to Google in 2010 as an “implied” backlink, and since then have become a rather important element of SEO. With the integration of AI into search, the context of these mentions has been integrated into that calculation, and gAI search uses both positive and negative mentions in its results. Google are saying that inauthentic mentions will be treated like spammy backlinks.

Considering the difference in effort between an inauthentic mention and an authentic mention, this “myth” is essentially saying, “Don’t use bots to flood Reddit with brand mentions.” Which, honestly, is good advice, mostly because the negative publicity you’d get from thousands of disgusted Redditors would do you more harm than good anyway.

The main warning to take from this is that third parties offering to sell you a lot of mentions on social media, or automated comments mentioning your brand on YouTube, are likely to do you more harm than good. This sort of behaviour can, and likely will, result in a manual action against your website.

Overfocusing on structured data

SEO professionals have long loved structured data. It’s important information about you and your services hidden in the background of your website. There are lots of reasons to include structured data, but AI is not one of those reasons. Never has been, probably never will be.