AI Search Myths, Checked: How Do You Actually Show Up?

You may have seen the question "Are you ranking on ChatGPT?" It's the idea behind two newish terms: GEO (generative engine optimisation) and AEO (answer engine optimisation). Both mean trying to get your business mentioned or recommended in the answers AI tools give, such as ChatGPT, Gemini or Google's AI Overviews. The pitch is usually the same: the old rules are dead, there's a secret formula, and you need it before everyone else does. It's a lot like fad diets. New name, big promises, and underneath it the boring fundamentals still do the work.

There are several myths I'm seeing around AI search results, so here are seven of them, checked against what's actually been said. I've rated each one Busted or Mostly false. At the end I've also listed what isn't a myth.

How does an AI actually find things?

Think about how you'd answer if a friend asked you to find a good plumber. You wouldn't read the whole internet. You'd search, scan what came up, check the star ratings and the reviews, and maybe ring a couple.

AI tools work a lot like that, in two steps:

1. Get found. They search first, so they can only pick from what turns up. Google says this about its own AI features: they're "rooted in our core Search ranking and quality systems" and pull relevant pages from the Search index. [1] Other tools reportedly use a mix of their own indexes and other companies' search engines, and those arrangements can change. [2]
2. Get chosen. From what turns up, they pick using signals of trust, the same way you'd use stars and reviews.

Nobody outside these companies knows exactly how step two works, and the tools differ from each other. Keep that in mind for the rest of this post.

Myth 1: "AI search is a whole new discipline"

Mostly false, but even experts disagree.

Google's guide to AI features asks "Is SEO still relevant for generative AI search?" and answers "In short, yes!" It adds that "optimizing for generative AI search is optimizing for the search experience, and thus still SEO." [1]

Not everyone agrees. A January 2026 piece in Search Engine Journal says "there is no distinct GEO playbook", but also that there are "measurable differences in retrieval, citation, and answer synthesis". [3] The sensible reading is that it's mostly SEO with some new wrinkles, not a separate discipline you need to buy.

Myth 2: "You need an llms.txt file or special markup"

Busted, for Google.

Google says: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them." It also says there's "no special schema.org markup you need to add". [1]

Two fair caveats. Google says it's fine to make an llms.txt file for other services that use them, since it "will neither harm nor help" your Google visibility. And structured data is still worth using for ordinary rich results. [1] Outside Google, a Search Engine Land writer found no evidence that llms.txt increases AI citations, [4] and Search Engine Journal noted that no AI search engine currently uses it. [3]

Myth 3: "Chop your content into small pieces and write it a special way for AI"

Busted, for Google.

Google says "there's no requirement to break your content into tiny pieces for AI to better understand it", and "there's no ideal page length". It also says "you don't need to write in a specific way just for generative AI search", because its systems understand synonyms and general meanings. [1]

Clear headings and short paragraphs are still good for readers. That's a different thing from writing for a machine.

Myth 4: "The more your brand is mentioned, the higher AI will rank you, so buy some mentions"

Mostly false.

The thinking goes like this: if AI tools draw on what's said about a business across the web, then more mentions must mean more visibility, so why not create or buy some? It's the same logic as buying followers. The number goes up, but nobody trusting you has changed.

The key word in Google's guide is inauthentic. Google says its AI features can show what's being said about products and services across the web, including blogs, videos and forum discussions. But it says "seeking inauthentic 'mentions' across the web isn't as helpful as it might seem", because its ranking systems focus on high-quality content while other systems block spam. [1] Real mentions come from being worth mentioning.

That's Google, though. Nobody has shown how other AI tools weigh mentions, and I haven't found good evidence that piling up more of them pushes you higher anywhere.

Myth 5: "Backlinks don't matter any more, because AI gives the answer first"

Mostly false, though the details are thin.

This is the one I saw on Facebook, and it skips a step. Go back to the plumber. The AI still has to decide which pages are worth using as its sources, and for Google's AI features that decision comes from the same core ranking systems as normal search. [1] Whatever those systems use to judge a page is still in play, and links from other sites have long been one of the ways search engines judge trust.

What we can say:

• Google's guide doesn't say anything about backlinks, in either direction, so I can't point to Google for this one.
• Google still polices links. It says its systems can detect sites buying links and sites that exist to pass links on, and that the credit from those links is lost. [8] That suggests links still carry some weight.
• Google's John Mueller said in 2021 that the total number of links "doesn't matter at all", but that one really good link from a relevant site can be "a really important sign". [9]
• Microsoft reportedly lists backlinks among its SEO starting points. [3]
• A Semrush-run study of 1,000 domains found that sites with stronger authority scores and more unique linking domains tended to get mentioned more by AI tools. It was a correlation, it used a tool seller's own authority score, and the author called it "a first look". [5]

What we can't say is how much links count for AI tools specifically, or that they're the only thing that counts. Reviews, content quality and reputation matter too. So the honest version is that backlinks still look like one of the signals: not dead, and not magic. Be wary of anyone who's certain either way.

Myth 6: "If AI mentions you, the visitors will follow"

Busted.

In Pew Research's March 2025 data on 900 US adults, users clicked a traditional result in 8% of visits when Google showed an AI summary, against 15% when it didn't. Only 1% of visits included a click on a link inside the summary. [6] And a visit isn't a sale: typical ecommerce conversion rates sit around 2.3% to 2.7%. [7]

Being shown, clicked, visited and bought from is a chain that leaks at every step. That's why I'd measure enquiries and sales, not mentions.

Myth 7: "Just change the date and the page counts as fresh"

Mostly false.

Freshness itself isn't the myth. A Search Engine Land writer found it had the strongest support of the three practices he looked at, including studies and the fact that some AI tools treat freshness as a signal when deciding whether to search the web. [4] The myth is the shortcut: changing the date on a page without changing anything else. The same writer advises against it, saying Google keeps past versions and can detect it. That's his view, not something in Google's guide, which gives no advice on refreshing content, so treat it as a strong warning and not a proven rule. The test I'd use is whether there's something new to say. A real update means something changed: a new guideline, a new study, a correction. This post, for instance, reflects Google's guide as it stood when it was last updated on 10 July 2026. Getting a page looked at again isn't the same as getting it ranked either. Google's guide says that even a page that meets all the requirements isn't guaranteed to be crawled, indexed or served. [1] If nothing has changed, leave the date alone.

What does work?

Mostly the boring things, and Google's guide is blunt about them:

• Make something worth quoting. Google says unique, valuable content will "likely influence your website's presence in generative AI search in the long run more than any of the other suggestions in this guide". It warns against content that others "have already said, or could easily be produced by a generative AI model". [1]
• Be eligible. A page must be indexed and eligible to show with a snippet. Even then, Google says indexing and serving "aren't guaranteed". [1]
• Look after your local details. For local businesses and shops, Google says Google Business Profiles and Merchant Center can help you show up in AI responses. [1] That's the "get chosen" step for a plumber.
• Keep time-sensitive pages genuinely up to date. Of the practices in the Search Engine Land piece, freshness had the strongest support, [4] though that evidence comes from specific tools, not Google's guide.
• Measure it. Search Console has a Generative AI performance report. [1]

And a line I like: "plenty of content thrives in Google Search (including generative AI experiences) without any overt SEO at all, and you don't need to accomplish everything in this guide in order to succeed on Google Search." [1]

Where this leaves you

Be wary of any pitch that sounds like a fad diet. Google's own guide says so too: "No third-party tool has access to our internal ranking or AI systems." [1] Get found, then get chosen. Neither needs a secret, but both take real work.

The next set of myths will look at the services that sell all this, and whether they're worth the money.

Sources

1. Google Search Central, Optimizing your website for generative AI features on Google Search, last updated 10 July 2026. All Google quotes in this post are from this page.
2. Reports from 2024 and 2025. The Claude and ChatGPT details were not confirmed by the companies in these articles, and Perplexity's is the company's own claim. TechCrunch, Anthropic appears to be using Brave to power web search for its Claude chatbot (March 2025); Yoast, What is ChatGPT Search (and how does it use Bing data)? (November 2024); VentureBeat, Perplexity launches massive search API to take on Google's dominance (September 2025).
3. Search Engine Journal (Roger Montti), A little clarity on SEO, GEO and AEO, 21 January 2026, which also reports Microsoft's October 2025 guidance.
4. Search Engine Land (Philipp Götza), GEO myths and lies. The page shows no publication date.
5. Search Engine Journal (Kevin Indig, with Semrush data), How AI really weighs your links: analysis of 35,000 datapoints, 1 October 2025.
6. Pew Research Center (22 July 2025), as summarised by PPC Land. Based on March 2025 browsing data from 900 US adults.
7. Blend Commerce, Ecommerce conversion rate benchmarks 2026, citing IRP Commerce, Dynamic Yield and Shogun.
8. Search Engine Land, Google December 2022 link spam update done rolling out.
9. Search Engine Journal, Google's John Mueller: total number of backlinks doesn't matter, from a February 2021 Google hangout.