How to Rank in AI Search Engines (AEO)

Google isn't the only search engine anymore. When someone asks "which security scanner is right for a solo developer?", they're as likely to ask ChatGPT or Perplexity as to type it into Google — and the AI answers with a summary plus citations, not a list of links. If you're not in the citations, you're invisible.

AEO (Answer Engine Optimization) is how you get quoted. The good news: it overlaps with good SEO, and you can build most of it in an afternoon.

#What AI engines actually read

Before optimizing, understand the input side. AI answer engines do three things with your site:

1. Crawl it — through robots.txt, sitemaps, and links, just like Google
2. Render it — JavaScript-heavy pages are evaluated; content injected by client-side JS is indexed too
3. Extract from it — structured data (JSON-LD), headings, and direct-answer text get pulled into the model's context

That third step is where AEO lives. The model isn't ranking your page; it's deciding whether your page answers the question in a quotable form.

#The 7-point AEO checklist

#1. Let the AI bots in

Your robots.txt should explicitly allow the big answer-engine crawlers:

User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

Blocking these is the single most common way sites accidentally disappear from AI answers. Also ship an llms.txt file — it's the emerging machine-readable index (spec: llmstxt.org) that helps LLMs find your best content. For the underlying standard governing AI crawler access, see RFC 9309.

#2. One question, one direct answer

Structure each page around one primary question. Within the first 100 words, answer it in a standalone sentence someone could quote. Don't bury the answer behind an intro paragraph, a joke, or a product pitch.

#3. Use real headings (H2/H3), not styled divs

AI extraction relies heavily on heading structure to segment content. Use semantic <h1> → <h2> → <h3> with descriptive text — "## How AEO is different from SEO" beats "## Deep dive" every time.

#4. Ship structured data

JSON-LD is the machine-readable version of your page's meaning. The highest-leverage schemas:

• Article / BlogPosting — headline, dates, author, publisher
• FAQPage — question/answer pairs that AI assistants quote directly
• Organization — the "who is behind this" trust signal

Here's a minimal BlogPosting block:

{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "How to Rank in AI Search Engines (AEO)",
  "datePublished": "2026-08-01",
  "author": { "@type": "Person", "name": "Vergate Team" },
  "publisher": { "@type": "Organization", "name": "Vergate" }
}

#5. Write for extraction, not just readers

Lists, tables, and short paragraphs extract cleanly. Definitions and step-by-step instructions ("first… then… finally…") are the highest-citation formats in AI answers. Bold the key term in each section — models weigh it when picking the quotable sentence.

#6. Answer the question people actually ask

AEO rewards conversational, question-shaped content. If your page targets "which security scanner for indie devs", include the FAQ-style phrasing: "What is the best security scanner for solo developers?" — with a definitive answer. This is what FAQPage schema formalizes.

#7. Verify with a real AI-engine check

Don't guess — measure. Run an AEO analysis on your page and check the checks yourself: llms.txt present? AI bots allowed? JSON-LD valid? Single H1? Structured data for the answer? Vergate's AEO mode checks all 26 of these in seconds and tells you exactly what to fix.

#What this looks like in practice

This very site is the dogfood. Every post on this blog has:

• A single H1 from the title, semantic H2/H3 structure below it
• BlogPosting JSON-LD with author, dates, and word count
• FAQPage JSON-LD where the post answers real questions
• An llms.txt index and a sitemap that includes the post
• An RSS feed so syndicators and AI aggregators can pick it up

That's ~2 hours of setup once, then every post inherits it. The checklist above is the same checklist we scan with.

Ranking in AI is not a different game from ranking in Google — it's the same game with an extra extractor reading the same cards.

#Start now

Three things this week:

1. Add the AI-bot User-agent blocks to robots.txt and generate llms.txt
2. Add BlogPosting (and FAQPage where it fits) JSON-LD to your best pages
3. Run an AEO scan and fix the top five checks

And if you're shipping with AI, read about vibe coding security next — the two posts are the security and visibility halves of shipping an AI-first product.