By Mehdi HassanWhich AEO myths waste the most time?
The AEO myths that waste the most time are technical shortcuts sold as a separate discipline: llms.txt files, AI-only schema, chunked rewrites and separate AI versions of pages. Google's AI features documentation and its May 2026 guide say AI optimization is normal SEO, and independent tests back that up.
These myths persist because they're easy to sell and easy to ship. A file or a markup change feels like progress, while rewriting a vague homepage feels like hard work. Our answer engine optimization guide covers what does work. Here are the 12 tactics we see pitched most often, and what the evidence says.
Myth | What the evidence says | Do this instead |
|---|---|---|
llms.txt gets you cited | 97% of files got zero requests (Ahrefs) | Make sure crawlers can read the HTML |
AI-specific schema lifts citations | No lift in Ahrefs' 2026 controlled test | Keep standard schema as hygiene |
Chunk pages into tiny sections | Google advises against it | Write self-contained sections of normal length |
You must rank top 10 to be cited | Conductor: 57% of citations come from outside it | Answer the question better than anyone |
Backlinks are the main lever | Mentions correlated 0.664, backlinks 0.218 | Earn mentions on trusted sites |
Best-of lists win ChatGPT | Self-promotional lists: about 4% of citations | Publish honest comparisons |
Which technical AEO myths can you ignore?
You can ignore the claims that a special file, a special markup type or a special AI version of your site is required. None of them has evidence of moving citations, and Google says its AI features use the same systems as regular search.
Myth 1: you need an llms.txt file
Ahrefs found 97% of llms.txt files received zero requests, and Google lists the file as unnecessary for its AI features. It's harmless to add, but it isn't a strategy. If crawlers can't read your content, the fix is server-rendered HTML, not a text file.
Myth 2: AI-specific schema gets you cited
Ahrefs' controlled 2026 schema test found no lift: AI Overview citations fell 4.6% while AI Mode and ChatGPT stayed flat. Keep Article, Organization, FAQ and Breadcrumb markup for rich results and clarity. Don't expect any markup to win citations by itself.
Myth 3: you need separate AI versions of your pages
Google's guidance calls AI-specific rewrites unnecessary. Building a parallel set of pages for machines also risks duplicate content and pages that drift out of date. One clear page for people is the page engines want too. See Google's AI features documentation for its own wording.
Which content AEO myths hurt your pages?
The content myths that hurt pages are the ones that make them worse for people: chopping text into fragments, stuffing prompt phrases and hitting word counts. Engines quote clear, specific answers, and those come from good writing, not formatting tricks.
Myth 4: chunk everything into tiny pieces
Google advises against breaking content into tiny chunks for AI. What works is a self-contained section under each question-style heading, with a direct answer in the first sentence or two. AirOps reports more citations for sections of about 120–180 words, though it's a vendor heuristic, not a rule.
Myth 5: repeat the prompt wording everywhere
Keyword stuffing doesn't become useful because the keyword is now a prompt. Use the buyer's words in headings and the opening answer, then write naturally. Engines paraphrase anyway.
Myth 6: an FAQ block with schema guarantees citations
A visible FAQ with real buyer questions helps people and gives engines clean answers to lift. The schema alone does nothing for citations, as the Ahrefs test showed. Our guide to FAQ sections that get cited covers what to write.
Myth 7: longer pages always win
Length helps only when it adds answers. A page that covers the sub-questions buyers ask beats a short page, but padding a page to a target count adds nothing an engine would quote.
Which ranking AEO myths should you stop believing?
Stop believing that AI citations simply mirror the top 10, that backlinks are the main lever, and that you can game ChatGPT with lists or forum posts. The research points to answer quality and off-site reputation instead.
Myth 8: you must rank top 10 to be cited
Conductor found 57% of AI Overview citations come from outside the top 10, and BrightEdge puts the overlap near 17%. Methods differ, but rankings help less than they used to. Read how AI search engines choose sources for the full picture.
Myth 9: backlinks are what matter most
In Ahrefs' study of 75,000 brands, web mentions correlated 0.664 with AI Overview visibility, against 0.218 for backlinks. Links still matter for SEO, but being talked about on trusted sites matters more for AI answers.
Myth 10: self-promotional best-of lists win ChatGPT
Peec's analysis of 232,000 citations found self-promotional lists made up only about 4% of ChatGPT citations, and Seer reported listicle citations falling 30% month on month. Honest comparisons that admit trade-offs are a better bet.
Myth 11: seeding Reddit is a shortcut
Reddit's share of ChatGPT citations has swung sharply, and it behaves differently in Google than in ChatGPT. Real participation can help your reputation; planted posts are a fragile, risky tactic to build on.
Which measurement AEO myth misleads teams?
The measurement myth that misleads teams is that one check, or one tracker, tells you where you stand. AI answers vary from run to run and engine to engine, so a single snapshot is noise.
Myth 12: one prompt check shows your AI visibility
Semrush's AI Visibility Index found ChatGPT cites about 15 sources per answer against about 3 for Gemini, so visibility on one engine says little about another. Run a fixed set of 20–30 buyer prompts monthly across engines, log the results, and read the trend over a quarter. Combine it with a GA4 AI channel group and a "How did you hear about us?" field.
What actually moves AI citations?
What moves AI citations is the boring work: pages that answer buyer questions directly and specifically, content crawlers can read, a named author, fresh updates and mentions on sites engines trust. The money pages carry the most weight.
- Answer first. Put a standalone answer in the opening lines of each section. See how to structure content so AI engines quote it.
- Server-render the content. Many AI crawlers don't run JavaScript, so the answer must be in the HTML. Our guide to an AI-search-ready website covers the build.
- Show who wrote it. Named bylines and real author pages support trust signals.
- Keep it current. AirOps says over 70% of cited pages were updated within 12 months; see content freshness for AI search.
- Fix the homepage and pricing first. TripleDart found homepages carried 81.7% of AI-attributed leads across 28 B2B SaaS companies.
How do you spot an AEO pitch built on myths?
You spot a myth-based AEO pitch when the deliverables are files, markup and reports rather than rewritten pages and earned mentions. Ask what changes on your site that a buyer would notice.
Five questions to ask any AEO vendor
1. Which pages will you rewrite, and who writes the copy?
2. What evidence do you have that each tactic moves citations, beyond your own case studies?
3. How will you measure citations across ChatGPT, Perplexity, Gemini and Google, and how often?
4. What will you do about mentions on sites we don't own?
5. Which of your deliverables does Google's own AI guidance call unnecessary?
If most answers are about llms.txt, schema packages or AI-specific page versions, you're paying for the myths in this list.
