The Complete AEO Checklist: How to Get Your Website Cited by AI

Start with crawling and indexing

Google's AI answers are built from its normal search index. Its guide to generative AI features says plainly that optimizing for them is still SEO, because AI Overviews and AI Mode sit on top of the same core ranking and quality systems. A page that Google can't crawl, index, or rank has nothing to offer the AI layer.

So the first check is the unglamorous one: important pages return a 200, aren't blocked or noindexed, sit in your sitemap, and have internal links pointing at them. If you want the full picture of how that ranking layer works, we mapped it in how to rank your website in the AI era.

One distinction trips people up: robots.txt controls crawling, and noindex controls indexing. Blocking a page in robots.txt doesn't guarantee it disappears from results, and a crawler can't see a noindex tag on a page it isn't allowed to fetch.

For how those ranking systems came to work the way they do, see our history of Google's search algorithm.

Let the AI search bots in

Every major AI company runs more than one crawler, and they do different jobs. The one that decides whether you show up in ChatGPT search is OAI-SearchBot. GPTBot is the training crawler, and OpenAI treats the two settings as independent. Anthropic and Perplexity split their bots the same way. That means you can stay visible in AI search while opting out of training, if that's what you want.

A minimal robots.txt that allows the search bots looks like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

Then check the layer above robots.txt. CDNs and firewalls often block AI crawlers by default, so look in your server logs for these user agents and confirm they're getting 200 responses, not 403s or bot challenges.

The volume is worth the effort. Vercel counted a month of crawler requests across its network: GPTBot, Claude, AppleBot, and PerplexityBot together made a little over 28 percent as many fetches as Googlebot.

Put your content in the HTML, and make it fast

Googlebot renders JavaScript. Most AI crawlers don't. Vercel's analysis of crawler traffic found that the OpenAI, Anthropic, and Perplexity crawlers it measured fetched pages without executing JavaScript. If your pricing, product details, or article text only appear after client-side rendering, those crawlers see an empty shell.

The test takes a minute: open a key page, view the page source (not the inspector), and search for a sentence from the main content. If it isn't there, move that page to server-side rendering or static generation.

Speed belongs on the same check. Google's Core Web Vitals targets, measured on real visits at the 75th percentile, are the practical goal:

To get changes discovered faster, keep your XML sitemap accurate and use IndexNow to notify Bing and the other participating engines when a URL is published, updated, or removed. Copilot draws on Bing's index, so this helps AI visibility as well. Google doesn't use IndexNow; for Google, request a recrawl through URL Inspection in Search Console.

Answer the question early

State what the page answers, give the answer near the top, then follow with the reasoning and detail. Use headings that describe what each section covers, and tables or steps where they help. An answer engine looking for a quotable passage finds one quickly, and so does a human reader.

Don't take this too far. Google's guide says there's no requirement to break content into tiny pieces for AI and no ideal page length. Write the page at the length the topic needs.

Publish what AI can't copy

Google's guide asks for non-commodity content: things a reader couldn't get by asking a generic assistant. That means your own data, tests you ran, screenshots of real work, and opinions from people with a name and a track record. A page that restates what ten other pages already say gives an answer engine no reason to cite it over them.

There's research behind this too. The Princeton-led paper that coined the term Generative Engine Optimization found that adding statistics, quotations, and cited sources raised a page's visibility in generated answers by up to 40 percent, while keyword stuffing did little.

Before you write, match the page to what the searcher wants. A how-to question needs a guide, a comparison query needs an honest comparison, and a buying query needs a product or pricing page. Group related questions onto one page, and only split them when the intent really differs.

Show who wrote it, and review what AI drafts

Make it easy to see who is behind a page and how its claims were checked. That means author bios with relevant experience, links to primary sources, company and contact details, an editorial policy, and clear disclosure of commercial relationships. It matters most in health, finance, legal, and security topics, where a wrong answer costs the reader something.

Using AI to help write is fine. Google's guidance allows it as long as the result is accurate, useful, and reviewed, and it treats mass-produced low-value pages as scaled content abuse under its spam policies. The workable process is simple: research with traceable sources, draft, have someone who knows the subject edit and fact-check, then publish. Fewer good pages beat hundreds of generic ones.

Get the on-page details and structured data right

These are the elements a search engine or an answer engine reads first to work out what a page is about:

On structured data, use the types that describe what the page really is, such as Organization, Article, BreadcrumbList, Product, or LocalBusiness, and validate them before release. Google says schema isn't required for its AI features, so treat it as a way to be understood and to qualify for rich results. Never mark up ratings, reviews, or facts that aren't on the visible page.

Earn real mentions on other sites

What other sites say about you tracks AI visibility more closely than anything on your own site. Ahrefs studied 75,000 brands and measured how strongly different signals correlated with appearing in Google's AI Overviews:

These are correlations, not proof of cause. Still, the pattern is consistent: being talked about matters more than raw link count. The sources that produce mentions are reviews, digital PR, expert quotes in industry publications, and research that others reference.

The word that matters is earned. Google's guide warns that seeking inauthentic mentions isn't as helpful as it seems, and its spam systems target them the same way they target link schemes. Our guide to getting backlinks without outreach covers the kinds of assets that earn both links and mentions.

If you trade links with partner sites, read what Google's policies say about link exchange first.

Local businesses and online stores

If you serve a location, your Google Business Profile is a direct source for local answers. Keep the name, address, phone number, hours, and services accurate, and make sure they match your website and other listings. Collect honest reviews and reply to them.

If you sell products, AI shopping answers depend on your product data. Keep prices, availability, identifiers, shipping, and returns accurate in Merchant Center and identical to what the product page shows. Product pages need to be crawlable, with the key details in the HTML.

Make pages that AI agents can use

Being cited is one thing. People are also starting to ask browser agents to compare products, fill in forms, and book things for them. Google's web.dev guidance explains that agents read a page through screenshots, the raw HTML, and the accessibility tree, so a site that works well with a screen reader usually works well with an agent.

None of this is a ranking factor. It decides whether an agent can finish a task on your site once it gets there. WebMCP, a proposed standard for exposing site actions to agents as structured tools, is worth watching if you have a real task to simplify, such as a quote request or a booking flow. It's still experimental, so it can wait until the rest of this list is done.

Track your AI citations

You no longer have to guess. Google Search Console's Generative AI performance report shows impressions for links to your site in AI Overviews and AI Mode, by page, country, device, and date. Google rolled it out to all sites worldwide by August 31, 2026.

Bing Webmaster Tools has an AI Performance report that counts how often your pages are cited in Copilot and Bing's AI summaries, along with the grounding queries behind them.

ChatGPT, Claude, and Perplexity don't offer a report like that yet. For those, keep a fixed list of 20 or so questions your customers ask, run them monthly, and log whether you were mentioned and which URLs were cited. One answer proves nothing. The trend over several months does.

In your analytics, separate visits referred by AI tools from ordinary organic search, and judge both by what they lead to: qualified leads, trials, and sales. An AI impression isn't a click, and a click isn't a customer. Be careful with third-party AI visibility scores too. Those tools can't see inside any engine's ranking system, so check their claims against your own data.

What you can skip

Four popular AEO tactics that Google's own guide addresses directly:

A 90-day plan

The order matters more than the speed. Fix access first, then content, then authority:

The full AEO checklist

Everything above, on one list you can work through:

AEO isn't a new discipline. It's a sound site that AI crawlers can read, saying something worth quoting, that other people already talk about.