"Does my website need an llms.txt file?" is a question more and more small business owners are asking. Usually it comes right after they have received a proposal that includes one, alongside a rewrite of their content "for AI" and a monthly visibility score.
This article is part of my small business guide to SEO.
The short answer is no, you do not need an llms.txt file, and several other widely repeated AI SEO claims do not stand up either. Below are eight of the most common myths, each checked against Google's published guidance, independent research and my own test of ChatGPT's local recommendations.
What is llms.txt?
llms.txt is a proposed standard: a plain text (Markdown) file placed at the root of a website, for example yoursite.co.uk/llms.txt. The idea is that it gives large language models (LLMs) a short summary of the site and links to its most important pages, a bit like a sitemap written for AI. If you are not sure what a sitemap is, my guide on what a sitemap is explains it.
It was proposed in 2024 and quickly became a staple of AI SEO checklists. Some SEO plugins, including SEOPress, can now generate one automatically.
The question that matters for your business is simple: does it help you get found?
Myth 1: You need an llms.txt file to appear in AI answers
The claim: Without llms.txt, AI tools cannot properly understand or recommend your business.
The reality: No major AI company has confirmed it uses llms.txt to decide what to cite or recommend. Google's guidance on its AI search features says you "don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search", and that such files "neither harm nor help your site's visibility or rankings in Google Search."
Google's John Mueller put it more bluntly in 2025. He said none of the AI services had said they use llms.txt, that server logs showed they did not even check for it, and that it was "comparable to the keywords meta tag", a tag search engines stopped using many years ago. He also pointed out the obvious flaw: AI tools can already read your actual pages, so a separate summary written by the site owner adds little and could be abused.
What to do: If your SEO plugin can generate an llms.txt file for free, there is no harm in switching it on. Just do not pay for one, and do not expect it to change anything.
Myth 2: You need to rewrite your content into "chunks" for AI
The claim: AI tools extract passages, so you need to break every page into short, self-contained sections, or rewrite everything as questions and answers.
The reality: Google's guidance says: "There's no requirement to break your content into tiny pieces for AI to better understand it." It also says you "don't need to write in a specific way just for generative AI search", because AI systems understand synonyms and intent.
That does not mean structure is useless. Clear headings, direct answers near the top of a section and specific detail make content easier for anyone to use, human or machine. That is simply good writing, and I cover it in how to optimise website content for SEO.
What to do: Write clearly for your customers. Do not pay for a rewrite whose only justification is an "AI format".
Myth 3: You need special schema markup for AI
The claim: AI tools rely on new or special structured data, so your site needs AI-specific markup.
The reality: Google's guidance says no new markup is needed to appear in its AI features. Standard schema (LocalBusiness, Service, Product, FAQ, Review) remains useful, as it has been for years, because it helps search engines understand your pages.
What to do: Make sure your site has sensible, accurate standard schema. That is a one-off technical SEO task, not an ongoing AI service.
Myth 4: Google rankings do not matter for AI search any more
The claim: AI tools pick sources in a completely different way, so your Google rankings are irrelevant.
The reality: It depends on the type of search. For broad, informational questions, the overlap between Google's top results and AI citations is real but partial: Ahrefs' March 2026 analysis found 38% of AI Overview citations came from Google's top 10.
For local searches, the overlap in my test was much stronger. When I ran 10 local searches through ChatGPT and compared them with Google:
- ChatGPT recommended 85% of the businesses in Google's map results.
- The star ratings and review counts ChatGPT displayed matched Google's exactly in 66 of 67 cases.
What to do: If you are a local business, your Google Business Profile and map rankings are your AI strategy. If you publish broader content, rankings still matter, but so does being mentioned and cited around the web.
Myth 5: SEO is dead because people use ChatGPT now
The claim: Search is moving to AI, so traditional SEO is a waste of money.
The reality: AI tools are growing quickly, but from a small base. Similarweb data for January 2026 put AI platforms at roughly 0.15 to 0.25% of global web traffic, while Google organic search accounted for around 48.5%. And as the sections above show, AI tools lean heavily on the same signals SEO works on.
What to do: Keep investing in SEO. The reasons are laid out in why your business needs SEO, and none of them have gone away.
Myth 6: You can track your "AI ranking" like a Google ranking
The claim: A monthly AI visibility score or "share of voice" percentage shows exactly how you are performing in ChatGPT.
The reality: AI answers vary each time you ask. In my test, I ran each search through ChatGPT three times, and only 40% of the businesses it recommended appeared in all three runs. For one search, ChatGPT named 30 different businesses and only one appeared every time. A score built on answers that shift that much has a large margin of error.
What to do: Search for your key services in ChatGPT yourself a few times each month, and keep an eye on referral traffic from chatgpt.com in Google Analytics. Measure enquiries, not mentions. If your SEO provider tracks ChatGPT visibility for you (I can set this up for my clients), judge it on the trend over several months rather than any single score.
Myth 7: Blocking AI bots does not affect your visibility
The claim: AI crawlers are just scrapers, so it makes no difference whether your site blocks them.
The reality: There are different kinds of AI crawler, and blocking the wrong one can hide you from AI answers.
| Crawler type | Examples | What blocking it does |
|---|---|---|
| Training crawlers | GPTBot, ClaudeBot, Google-Extended | Stops your content being used to train AI models |
| Search and answer crawlers | OAI-SearchBot, Claude-SearchBot, PerplexityBot | Stops AI tools reading your pages when answering questions |
| Google's main crawler | Googlebot | Removes you from Google Search, including AI Overviews |
Some hosting and security services, including Cloudflare, block AI training crawlers by default. That is a reasonable choice for many businesses. The risk is a blanket rule that also blocks the search crawlers, which quietly removes your site from the pages AI tools can read.
What to do: Check your robots.txt file and any firewall or bot protection settings, or ask whoever manages your site to do it. This belongs in any proper technical audit, and my guide on how to perform an SEO audit walks through the rest of the checks.
Myth 8: GEO is a completely separate discipline from SEO
The claim: Optimising for AI requires different skills, different tactics and a different (additional) budget.
The reality: Google says optimising for its generative AI features "is optimizing for the search experience, and thus still SEO." Google's Danny Sullivan has said "good SEO is good GEO." Some things have shifted in emphasis, such as brand mentions and crawler access, but they are all things SEO already covers. I set out the full case in GEO vs SEO.
What to do: Before paying extra for GEO, ask what the provider will do that your SEO does not. My breakdown of what a GEO add-on actually buys goes through the usual deliverables line by line.
AI SEO myths at a glance
| Myth | Verdict |
|---|---|
| You need an llms.txt file | False. No confirmed effect |
| You need to chunk content for AI | False, according to Google |
| You need special AI schema | False. Standard schema is enough |
| Google rankings no longer matter | False, especially for local businesses |
| SEO is dead | False. AI traffic is still a small share |
| AI rankings can be tracked precisely | False. Answers vary run to run |
| Blocking AI bots makes no difference | False. Blocking search crawlers hides you |
| GEO is a separate discipline | Mostly false. It is SEO with new emphasis |
What actually helps you appear in AI answers
If you want to be recommended by ChatGPT, Google's AI Overviews and whatever comes next, focus on the fundamentals that showed up in the evidence:
- A complete Google Business Profile with the right categories and service areas.
- Genuine reviews, because ChatGPT displayed Google ratings exactly.
- Accurate listings on relevant directories and trade bodies, which were cited for 9 of the 10 searches in my test.
- A crawlable, fast website with clear service and location pages.
- Real mentions of your business from local press, partners, suppliers and industry bodies.
And if you are weighing up a proposal that leads with llms.txt files and AI visibility scores, read what GEO agencies are really selling you before you sign.
Conclusion
Most of the AI SEO myths doing the rounds share one feature: they turn something free, basic or unproven into something you pay for monthly. llms.txt has no confirmed effect, Google says you do not need to rewrite or chunk content, and AI answers are too unstable to track like rankings. What does make a difference is the same work that gets you found in Google.
Every SoNick SEO package includes technical SEO, Bing Webmaster Tools set-up, Google Business Profile management and monthly reporting as standard, and I can set up ChatGPT visibility tracking too. There is no separate GEO retainer. If you want to know whether anything on your site is holding you back in Google or AI search, run a free SEO audit. It takes under a minute.
