The way people search is changing. Is your site ready?
For the past two decades, SEO (Search Engine Optimization) was the undisputed king of web visibility. Rank on Google's first page and you win traffic. Build backlinks, stuff keywords, optimize meta tags — the playbook was well understood.
Then something shifted.
People stopped clicking links. They started asking questions — and getting answers directly from ChatGPT, Perplexity, Claude, and Google AI Overviews. No click required. No visit to your site. Just an answer, synthesized from content scraped from across the web.
If your site isn't structured for AI agents, you don't just rank lower. You don't exist.
That's where AEO — Answer Engine Optimization — comes in.
What Is SEO? (A Quick Recap)
SEO is the practice of optimizing your website so that traditional search engines (Google, Bing, DuckDuckGo) rank it highly in results pages.
The core mechanics:
Crawlability — bots like Googlebot read your HTML
Relevance — keyword matching between your content and user queries
Authority — backlinks signal trustworthiness
Technical signals — page speed, mobile-friendliness, structured data (JSON-LD, Schema.org)
UX signals — click-through rate, dwell time, bounce rate
The end goal: appear on page 1, earn the click, get the visit.
What Is AEO?
AEO is the practice of structuring your content so that AI-powered answer engines can read, understand, and cite it directly in their responses.
The core mechanics are different:
Content negotiation — AI agents send Accept: text/markdown or a known bot User-Agent. Your server must respond with clean markdown, not HTML full of scripts and ads.
Machine-readable format — Markdown is what LLMs prefer. It strips away visual noise and leaves pure content.
Discovery signals — Link rel="alternate" headers, llms.txt files, .md twin URLs tell AI agents where to find the right content.
Correct HTTP headers — X-Robots-Tag: noindex, Vary: Accept, X-Markdown-Tokens, X-AEO-Version signal to the AI pipeline how to handle your content.
The end goal: be cited in an AI-generated answer, without requiring a click at all.
SEO vs AEO — Side by Side
| Dimension |
SEO |
AEO |
| Target |
Google, Bing, DuckDuckGo |
ChatGPT, Perplexity, Claude, Gemini, AI Overviews |
| Client type |
HTML-rendering crawler |
LLM inference pipeline |
| Content format |
HTML + structured data |
Plain Markdown |
| Discovery |
Sitemap, robots.txt, backlinks |
Link rel=alternate, llms.txt, .md URLs |
| Key headers |
Cache-Control, canonical |
Content-Type: text/markdown, Vary: Accept, X-Markdown-Tokens |
| Success metric |
Ranking position, organic clicks |
Citations in AI answers, zero-click answers |
| What hurts you |
Thin content, poor backlinks, slow pages |
Serving HTML to bots, missing headers, no markdown twin |
| Spec body |
Google's guidelines (de facto) |
AEO Spec v1.0 (RFC-based, community-driven) |
| Protocol |
Implicit (Google's algorithm) |
Explicit HTTP content negotiation (RFC 7231) |
Why Can't I Just Do SEO and Ignore AEO?
Good question. Here's why SEO alone is no longer enough:
1. AI answers are eating clicks
Google's own AI Overviews now appear above organic results for hundreds of millions of queries. Perplexity, ChatGPT Search, and Claude all synthesize answers from web content — without sending traffic to the source.
If you're not cited, you don't exist in that answer. Ranking #1 on Google doesn't help if the AI ignores your HTML.
2. AI crawlers behave differently from Googlebot
Googlebot renders JavaScript. It understands your React SPA. It indexes your HTML.
AI crawlers like GPTBot, ClaudeBot, and PerplexityBot often don't. They send Accept: text/markdown, or they identify themselves with a known bot User-Agent, and they expect a clean text response — not a hydrated React component tree with three ad scripts loading.
If your server responds with text/html to Accept: text/markdown, you've already failed the most basic AEO check.
3. LLMs train on and retrieve markdown
The pretraining data that powers GPT-4, Claude, and Gemini is overwhelmingly plain text and markdown — not HTML. When these models retrieve content at inference time (RAG, search grounding), they prefer the same format.
Serving markdown is not just polite — it increases the probability your content is correctly parsed and cited.
4. Zero-click is the new normal
Studies already show that a significant and growing portion of searches end without a click. For informational queries — "what is X", "how does Y work", "best Z for students" — AI gives the answer inline.
Your SEO strategy optimizes for clicks. AEO optimizes for being the answer — whether or not anyone clicks through.
What Does AEO Actually Require?
The AEO Spec v1.0 defines six things a conformant server must do:
1. Content Negotiation
Honor the Accept header. If a client sends Accept: text/markdown, respond with markdown. The spec builds on RFC 7231 §5.3.2 — a 10-year-old IETF standard.
2. Markdown Twins
Every HTML page should have a .md counterpart (e.g. /about → /about.md) containing the same content as clean markdown.
Content-Type: text/markdown; charset=utf-8
X-Markdown-Tokens: 412 ← token count estimate
X-Robots-Tag: noindex ← prevent duplicate indexing by traditional crawlers
Vary: Accept ← tell caches the response varies by Accept header
X-AEO-Version: 1.0
X-Content-Type-Options: nosniff
5. Discovery (on HTML responses)
Link: </about.md>; rel="alternate"; type="text/markdown"
Vary: Accept
6. AI Bot Detection
Serve markdown automatically when the User-Agent matches a known AI crawler — even if they don't send Accept: text/markdown.
Do SEO and AEO Conflict?
No. They are complementary.
Good SEO practices — quality content, clear structure, internal linking, fast pages — are also good AEO practices. The markdown you write for AEO is often the clearest, most readable version of your content, which is exactly what traditional crawlers want too.
The key additions AEO requires are infrastructure (content negotiation, proper headers) and format (markdown twins). Your content strategy stays the same.
Think of it this way:
You need both.
How to Check Your AEO Score
The Dualmark Playground runs a free 0–125 readiness check against the AEO Spec in seconds. It tests all 11 core checks across three categories:
md.* — Are your markdown twin responses correct?
html.* — Do your HTML responses carry discovery signals?
negotiation.* — Does your server actually do content negotiation?
Scoring levels:
| Score |
Level |
| 0–39 |
Below Basic |
| 40–74 |
Basic |
| 75–99 |
Standard |
| 100–125 |
Advanced |
Most sites score below 40 today. The opportunity is wide open.
The Bottom Line
SEO got you ranked. AEO gets you cited.
As AI-powered answer engines continue to absorb search intent, the sites that win won't necessarily be the ones with the most backlinks or the best-optimized title tags. They'll be the ones whose content AI agents can actually read, parse, and trust.
AEO isn't a replacement for SEO. It's the next layer.
The good news: the spec is built on open HTTP standards, the test suite is free, and the implementation is a few files of edge middleware away from any modern deployment.
There's no reason to wait.
Test your site at dualmark.dev/play — takes 10 seconds.
AEO Spec v1.0 reference: dualmark.dev/docs/spec/overview
