AEO Guide

What is AEO? The SEO for AI Agents Explained

Search engine optimization taught us to think like Google's crawlers. Now there's a new audience to optimize for: AI agents. ChatGPT answers 100 million daily queries. Perplexity has crossed 15 million daily active users. Claude is embedded in thousands of developer workflows. When these systems answer "what's the best invoicing tool?" or "which payment processor should I use?", they're deciding which companies exist — and which don't.

Answer Engine Optimization (AEO) is the discipline of making your site legible, trustworthy, and discoverable to AI agents. It's not a replacement for SEO. It's what comes next.

What is AEO (Answer Engine Optimization)?

AEO is the practice of structuring your web presence so that AI systems can accurately understand what you offer and confidently recommend you in response to user queries. Where traditional SEO targets keyword rankings in a results list, AEO targets synthesis — being included in the answer an AI constructs, not just a link in a list.

The distinction matters because AI agents don't paginate. They don't show you 10 blue links and let you choose. They form an opinion and state it. If your site is technically inaccessible to their crawlers, semantically unclear in its structure, or absent from their training data signals — you're invisible. No ranking. No impression. Just absence.

AEO covers the full surface area of AI discoverability: structured data that signals entity relationships, crawl permissions that explicitly allow or deny AI bots, machine-readable site maps for LLMs, and content architecture that makes your value proposition parseable in under 500 tokens.

Why It Matters Now

The shift is already happening. A 2024 study by SparkToro found that zero-click searches (where users get their answer without visiting a site) reached 58.5% of all Google searches. For AI-native tools like Perplexity, the number is closer to 90%. If you're not the source being cited, you're not getting the visit.

More critically: AI agents aren't just answering informational queries. They're handling transactional ones. "Find me a project management tool that integrates with Slack and costs under $20/seat" — that's a buying decision being delegated to an AI system. The companies with high answer engine optimization scores are the ones getting recommended. Everyone else is paying for ads to compete with a system that increasingly ignores ads.

The compounding effect: AI systems learn from usage feedback. When Claude or ChatGPT recommends your product and users follow through, that signal reinforces future recommendations. Early AEO investment compounds like domain authority — but the window to build it is right now, before the space is saturated.

The 6 Factors That Determine Your AEO Score

After analyzing hundreds of sites with our AEO Checker, the signal breaks into six distinct factors. Each is independently measurable. Most production sites are mediocre on at least three of them.

🏗️
Structured Data
JSON-LD schemas (Organization, Product, FAQ, HowTo) give AI systems explicit entity relationships. Most sites have none, or only basic meta tags.
🤖
robots.txt Configuration
Do you explicitly allow or block AI crawlers? GPTBot, ClaudeBot, PerplexityBot each respect different directives. Silence ≠ permission in all cases.
📄
llms.txt
An emerging standard: a curated, machine-readable index of your site's most important pages. Lives at /llms.txt. Currently rare — early adopters gain a clear signal advantage.
📐
Content Structure
Clear H1/H2/H3 hierarchy, descriptive meta tags, Open Graph data. AI agents parse HTML; semantic clarity directly maps to answer quality.
🔌
Tool / API Descriptions
If you have a public API or integration, is it described in a way an AI agent can relay? MCP-compatible descriptions and OpenAPI specs directly impact AI agent usability.
Performance & Accessibility
Slow sites get deprioritized by crawlers. JavaScript-heavy SPAs that render nothing server-side are effectively invisible to most AI crawlers today.

Real Data: We Tested Stripe, Anthropic, and OpenAI

To validate the scoring model, we ran the AEO Checker against three companies that should know better — they're either AI companies or deeply integrated with the AI ecosystem. The results are instructive.

Company
Score
Breakdown
Stripe
stripe.com
68/100
Best in class
Anthropic
anthropic.com
46/100
Moderate gaps
OpenAI
openai.com
41/100
Significant gaps

What separates Stripe from the others

Stripe's advantage is structural, not accidental. Their site has deep JSON-LD coverage (Organization, Product, BreadcrumbList across most pages), comprehensive Open Graph metadata, explicit robots.txt rules with named bot directives, and a server-rendered HTML stack that crawlers can actually parse. Their documentation is structured in a way that reads almost natively as a training corpus.

Anthropic and OpenAI's lower scores are surprising given their domain expertise, but make sense when you dig in. Both sites rely heavily on JavaScript-rendered content on key marketing pages. Neither has an llms.txt. Anthropic's robots.txt has a notable omission: it blocks several AI crawlers without providing a fetch endpoint for approved access. OpenAI's structured data is sparse outside of the blog — ironic for a company whose products consume structured data for breakfast.

The lesson: technical competence doesn't automatically translate to AEO awareness. This is a new discipline. Even the people building the AI systems haven't fully optimized for being consumed by them.

How to Check Your AEO Score

The AEO Checker analyzes any public URL across all six factors and returns a scored breakdown with specific, actionable findings. Enter your domain, wait about 10 seconds, and you get a report covering what's present, what's missing, and what the delta is worth in terms of AI discoverability.

The check is free, requires no signup, and runs entirely client-side against public endpoints — so you're auditing what AI crawlers actually see, not an authenticated internal view.

Actionable Checklist: Improve Your AEO Score

Most of the following can be implemented in a single afternoon. The ones marked with a star have the highest score impact per hour of work.

AEO Implementation Checklist
★ Add Organization JSON-LD schema to your homepage — name, url, description, logo, sameAs (social profiles). This is the single highest-leverage structured data item.
★ Create /llms.txt — a plain-text file listing your most important URLs with one-line descriptions. Format: # Site Name\n\n## Docs\n- /docs: Full documentation. Takes 20 minutes.
Review robots.txt for AI crawlers. Add explicit Allow: / directives for User-agent: GPTBot, ClaudeBot, PerplexityBot, Googlebot-Extended if you want to be indexed.
Add Product or SoftwareApplication schema to your product/pricing pages. Include description, offers, aggregateRating if you have reviews.
★ Add FAQ schema to your homepage or FAQ page. Question/answer pairs are directly consumed by AI systems for zero-click answers. Five good Q&As can meaningfully move your discoverability.
Audit your H1/H2 hierarchy. Every page should have exactly one H1. Section headers should describe the content, not be clever. AI parsers are literal.
Complete your meta and Open Graph tags. og:title, og:description, og:type, og:url on every indexable page. Thirty minutes with a template.
Ensure server-side rendering on key pages. If your marketing site is a React SPA with no SSR, AI crawlers see a blank page. Switch to Next.js, Astro, or add static HTML output.
If you have an API: write an OpenAPI spec and link it from your homepage. AI coding agents (Copilot, Cursor, Claude Code) directly consume these to generate integrations.
Add a sitemap.xml and reference it in robots.txt. Basic, but surprisingly absent from ~40% of sites we scan.

The Compounding Advantage

Traditional SEO results are measured in months. AEO operates on a different timeline: AI training cycles and crawler re-indexing are frequent and ongoing. A site that implements structured data, llms.txt, and explicit bot permissions this week may see improved representation in AI-generated answers within 4–8 weeks.

More importantly: the bar is still low. The companies we benchmarked — including AI companies — score under 70. For most industries, getting to 75+ puts you in the top percentile of AI-discoverable competitors. That window won't stay open indefinitely. The teams that treat AEO as a Q1 priority are building a moat that gets harder to close.

The mechanics are not complex. The structured data formats are documented. The llms.txt spec is a single page. The robots.txt directives are copy-paste. What's rare is prioritization. Fix your AEO score now, before your competitors realize they need to.

See how your site scores

Free analysis. No signup. Results in under 10 seconds. Find out exactly what's blocking your AI discoverability.

Check your AEO Score now 
Next read
How to Optimize Your Site for AI Agents →
7 concrete steps with code templates. Most can be done in 30 minutes.
AEO vs SEO: What's the Difference? →
A side-by-side comparison of what matters for search engines vs. AI agents.