Answer engine optimization is what SEO turns into when the search result is a paragraph instead of a list of links. ChatGPT, Claude, Perplexity, Gemini, and Google’s AI Overviews all compose answers and cite a handful of sources. Either your site is in that handful or it is invisible on that query.
This is the developer’s version of the topic: the structures, files, and markup that decide citability, with the code included. Everything here runs on this site, and it is the same checklist my SEO & AEO service implements for clients.
Short answer: AI engines cite pages that are fast, parseable, and answer-shaped. The work: put a direct 2 to 4 sentence answer at the top of every page, mirror real questions in your H2s, mark up entities and FAQs with schema, keep HTML clean and light, update important pages honestly at least every 6 months, and make your site readable to AI crawlers. None of it is exotic. Almost nobody does all of it.
Why AEO matters now: the numbers
Three findings shape the whole discipline:
- Queries with AI Overviews show roughly 37% lower organic click-through than the same queries without them, per Seer Interactive’s 2026 tracking. The old click volume is not coming back.
- Brands cited inside AI answers earn around 35% more clicks than uncited brands on those queries. The compressed pie pays the sources.
- Around 83% of AI citations go to pages updated within the last 12 months. Stale content is functionally invisible to answer engines.
So the strategy writes itself: bias toward being the citation, and keep everything fresh.
How AI engines choose what to cite
Reverse-engineering citations across engines shows consistent preferences. They favor pages that:
- Answer the question directly, near the top, in plain language
- Parse cleanly: semantic HTML, real headings, real tables, minimal JavaScript in the way
- Carry structured data that confirms who is saying it and why they are credible
- Are current, with honest signals of when they were last updated
- Agree with consensus while adding something specific: numbers, comparisons, first-hand detail
Everything below serves one of those five.
1. Answer-first content structure
The pattern AI engines lift most reliably: a question as a heading, then a 40 to 60 word direct answer, then the detail. On money pages I open with a verdict box, a blockquote holding the 2 to 4 sentence answer to the page’s core question. You are reading this pattern right now, and it is the single highest-leverage change on most sites.
Structure rules I apply to every page:
- H2s phrased as the questions people actually ask (check your Search Console queries)
- The direct answer in the first sentence under the heading, not the fifth
- Comparison content in real HTML tables, not styled divs
- Lists for anything enumerable: steps, criteria, options
2. Schema that actually matters
Schema is how you tell engines who is speaking. The set that earns its keep:
- Person + Organization with
sameAslinks to your real profiles: this is your entity, cited by name - Article with
authorand honestdateModifiedon every post - FAQPage wherever visible FAQs exist, mirroring the on-page text exactly
- Service + Offer with real prices on service pages
- BreadcrumbList sitewide
A minimal FAQPage block looks like this:
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [{
"@type": "Question",
"name": "Does llms.txt improve Google rankings?",
"acceptedAnswer": {
"@type": "Answer",
"text": "No. Google confirmed in June 2026 that llms.txt has no effect on Search. Claude, Perplexity, and agentic browsers do fetch it."
}
}]
}The rule that keeps schema safe: markup mirrors visible content. Schema describing things users cannot see is how you earn manual actions instead of citations.
3. Clean, light HTML
AI crawlers are worse at JavaScript than Googlebot. If your content only exists after a framework hydrates, some engines never see it. The fixes are architectural:
- Prefer static generation or server rendering for content pages
- Keep the answer in the initial HTML payload
- Use semantic elements:
<article>,<table>, real heading hierarchy - Watch JavaScript weight; a content page has no business shipping app bundles
This is why content stacks like Astro have an AEO edge by default, an argument I make with numbers in Webflow vs Astro.
4. llms.txt, with honest expectations
llms.txt is a markdown file at your site root that gives AI systems a curated map of your important pages. Honest status report: Google confirmed in June 2026 that it does nothing for Search or AI Overviews. But Claude, Perplexity, Cursor, and agentic browsers do fetch it, and it takes ten minutes:
# Your Name | yoursite.com
> One paragraph: who you are, what you do, who you serve.
## Services
- [Service name](https://yoursite.com/service): one-line description with real prices
## Guides
- [Best guide](https://yoursite.com/blog/guide): what question it answersShip it, keep it current, expect no rankings from it. This site’s version is at /llms.txt.
5. Let AI crawlers in (deliberately)
Citation requires access. The main user agents: GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, and Google-Extended (Gemini training). Blocking them is a legitimate choice for paywalled content businesses. For a services business, blocking them is refusing to be quoted in the rooms where buyers now ask questions. My robots.txt allows all of them on purpose:
User-agent: *
Allow: /
Sitemap: https://www.sanjeewa.works/sitemap-index.xmlDecide once, on purpose, per your business model. Do not block by default because a security checklist said so.
6. Freshness as a system
With 83% of citations going to recently updated pages, freshness is not a nice-to-have, it is table stakes. The system that works:
- Every important page gets touched at least every 6 months
- Updates are real: refreshed numbers, new sections, corrected claims, not fake date bumps
dateModifiedin schema and a visible “Updated [date]” label change together- A quarterly pass re-verifies prices and version numbers in old posts
7. Entity consistency everywhere
Engines cite entities they can resolve: same name, same role, same bio, same links across your site, LinkedIn, GitHub, and every profile. Your about page becomes the canonical entity page, every post carries the same author block, and sameAs ties it together. Inconsistency does not get penalized; it just quietly fails to accumulate.
Measuring it: the citation log
Positions and clicks still matter, but AEO gets its own scoreboard. Monthly, ask your 10 money questions in ChatGPT, Claude, Perplexity, and Gemini, and record who gets cited. It takes 30 minutes, builds a trend line no tool gives you yet, and tells you whether the work above is landing. Pair it with Search Console for the classic half of the picture.
The checklist
- Direct answer in the first screen of every money page
- H2s mirror real queries; 40 to 60 word answers underneath
- Person/Organization schema with sameAs; Article + author on posts
- FAQPage schema mirroring visible FAQs
- Content present in initial HTML; JavaScript minimal
- llms.txt shipped and current
- AI crawlers allowed (or blocked deliberately, per business model)
- Every important page updated within 6 months, honestly
- Monthly citation log across four engines
FAQ
Is AEO replacing SEO?
No. It layers on top. The same fundamentals (crawlability, speed, authority, good content) feed both, and most AEO work improves classic rankings too. What changes is the format of winning: a citation instead of only a ranked link.
Should I block AI crawlers to protect my content?
If your content is the product (courses, paywalled research), maybe. If your content exists to win clients, blocking crawlers removes you from the conversations where clients now ask for recommendations. It is a business-model decision, not a security default.
How long until citations show up?
Engines refresh at different speeds: Perplexity and ChatGPT can cite new pages within days to weeks, AI Overviews track Google’s index. Expect the citation log to show movement in 4 to 8 weeks for fresh, specific content, and remember the freshness clock keeps ticking after you win.
Can I do all this myself?
Yes, this guide is the complete list, and a technical founder can work through it in a few weekends. If you would rather hand it off, my SEO & AEO service audits your site against exactly this checklist ($500 to $1,500), implements the fixes from $1,250, and runs the freshness system from $550 per month.
